What does the AI Playbook for CRM AI: Prompts, Skills, Platform Guide & Agents include?

Buying, upgrading, or extending your CRM with AI? This guide cuts through the vendor noise so sales leaders, RevOps, and IT can make the right call.

Inside this playbook

Prompts • Skills • Connectors • Platforms • Agents • Buyer’s checklist

AI Playbook for CRM Teams

Q: How do CRM AI: Prompts, Skills, Platform Guide & Agents teams implement AI?

Set Up & Get Running. Begin with one low-risk review workflow using approved exports. Connect a platform only after an administrator has reviewed who may use it and what it may change.

A practical guide for buying, upgrading or extending a CRM with AI: prompts, reusable skills, researched platform and connection guidance, agent blueprints, and a buyer’s checklist.

100

AI Prompts

Claude-Ready Skills

Ways to Connect Your Work

CRM Platforms Compared

100AI Prompts

12Claude-Ready Skills

18Ways to Connect Your Work

12CRM Platforms Compared

Choose your next step

Buying or upgrading? Start with the platforms and the checklist. Already on a CRM? Start with a prompt, save a skill, then connect a narrow source.

→

+Suggest an edit

Start With Your Role

A CRM decision touches sales, operations, marketing, service and IT. Start with the work you own; keep record changes, customer contact and platform configuration with the person accountable for them.

Sales leader

Start with: pipeline health review
Save as a skill: approved inputs, output format, and stop rules
Connect: approved pipeline reports and stage definitions
First agent: pipeline hygiene reviewer
Measure: completeness, exceptions found, and reviewer changes

Revenue operations

Start with: CRM field completeness review
Save as a skill: approved inputs, output format, and stop rules
Connect: approved CRM exports and process documentation
First agent: data quality exception monitor
Measure: completeness, exceptions found, and reviewer changes

CRM administrator or IT

Start with: permission and integration inventory
Save as a skill: approved inputs, output format, and stop rules
Connect: approved configuration records and access matrix
First agent: configuration change reviewer
Measure: completeness, exceptions found, and reviewer changes

Marketing operations

Start with: lead routing and handoff review
Save as a skill: approved inputs, output format, and stop rules
Connect: approved campaign, form and routing rules
First agent: handoff gap organizer
Measure: completeness, exceptions found, and reviewer changes

Customer service lead

Start with: case handoff and escalation review
Save as a skill: approved inputs, output format, and stop rules
Connect: approved case records and escalation paths
First agent: escalation pattern organizer
Measure: completeness, exceptions found, and reviewer changes

Sales enablement

Start with: adoption and training gap review
Save as a skill: approved inputs, output format, and stop rules
Connect: approved usage reports and process guides
First agent: adoption digest
Measure: completeness, exceptions found, and reviewer changes

Data or analytics lead

Start with: reporting requirements specification
Save as a skill: approved inputs, output format, and stop rules
Connect: approved data dictionary and reporting definitions
First agent: report definition reviewer
Measure: completeness, exceptions found, and reviewer changes

Finance or deal desk

Start with: quote and approval process review
Save as a skill: approved inputs, output format, and stop rules
Connect: approved price book and approval thresholds
First agent: approval exception organizer
Measure: completeness, exceptions found, and reviewer changes

Partner or channel manager

Start with: partner pipeline review
Save as a skill: approved inputs, output format, and stop rules
Connect: approved partner records and joint pipeline
First agent: partner pipeline organizer
Measure: completeness, exceptions found, and reviewer changes

Executive sponsor

Start with: business case and adoption summary
Save as a skill: approved inputs, output format, and stop rules
Connect: approved baseline measures and project reporting
First agent: adoption and benefit summarizer
Measure: completeness, exceptions found, and reviewer changes

Start with the work, not the platform

The clearest way to compare CRMs is to run your own real work through each one and see what a person still has to fix.

→

+Suggest an edit

100 AI Prompts for CRM Teams

Ready-to-use prompts for pipeline, accounts, revenue operations, handoffs, lifecycle, reporting, configuration, win/loss, partners and CRM automation. Copy one, add approved CRM context, and have the record owner review the result.

12 prompts for Sales Managers, Account Executives, and Revenue Leaders. Use these to assess pipeline health, score deals, and prepare for forecast conversations.

Pipeline Health Review

You are a sales manager reviewing the current pipeline for forecast accuracy and deal quality. Pipeline data: [PASTE: Deal name | Account | Stage | Amount | Close date | Days in current stage | Last activity date | Owner] Analyze: 1. Stage distribution — are deals spread across stages or bottlenecked in one stage? 2. Stalled deals — any deal with no activity in >14 days or stuck in same stage >30 days; flag with days stalled 3. Close date realism — deals with close dates in the next 30 days; do stage and activity level support that timeline? 4. Pipeline coverage — total pipeline value ÷ quota; flag if below 3x coverage 5. At-risk deals — deals where close date has passed or activity has gone cold; recommend: pursue / reprice / close lost Output: Pipeline health report. Stalled deal list with recommended actions. Coverage ratio. Top 5 deals requiring manager attention this week.

Deal Scoring Assessment

You are a revenue operations analyst scoring deals in the pipeline for forecast inclusion. Deal data: [PASTE: Deal name | Stage | Amount | Close date | Champion identified? (yes/no) | Economic buyer engaged? (yes/no) | Compelling event? (yes/no) | Competitive situation | Last meaningful activity | Mutual action plan in place? (yes/no)] Score each deal on: 1. Engagement quality — are the right stakeholders involved and active? 2. Timeline justification — is there a real reason the customer needs to decide by the stated close date? 3. Competitive risk — is there an active competitor involved? What is our differentiation? 4. Process alignment — is there a mutual action plan or are we just waiting? 5. Overall forecast category: Commit (high confidence) / Best case (likely but not certain) / Pipeline (early stage) / At risk (stalled or at-risk) Output: Deal scoring table. Forecast category for each deal. Deals reclassified from Commit to At risk with reason. Total commit, best case, and pipeline values.

Deal Progression Analysis

You are a sales operations analyst reviewing deal progression rates. Pipeline data (last 6 months): [PASTE: Deal name | Start stage | End stage | Amount | Time in each stage (days) | Won/Lost/Open] Analyze: 1. Stage conversion rates — % of deals advancing from each stage to the next 2. Average time in each stage — where do deals slow down? 3. Drop-off stage — which stage has the highest deal loss rate? 4. Win rate by deal size — do larger deals win at the same rate as smaller ones? 5. Velocity — average days from first stage to close for won deals Output: Pipeline funnel analysis. Conversion rates by stage. Average time per stage. Drop-off analysis. Recommendations to improve conversion at the weakest stage.

Forecast Submission Prep

You are a sales manager preparing the weekly forecast submission. Pipeline data: [PASTE: Rep | Deal | Stage | Amount | Close date | Forecast category (commit/best case/pipeline) | Rep's confidence note] Build the forecast submission: 1. Roll-up by rep — each rep's commit, best case, and pipeline totals 2. Manager adjustments — deals you'd reclassify based on your knowledge; note reason 3. Coverage analysis — commit + best case as % of remaining quota for the period 4. Risk items — deals in commit that have warning signs; call out specifically 5. Upside items — deals in pipeline that could accelerate; note what would need to happen Output: Forecast submission table by rep. Manager-adjusted totals. Risk and upside narrative for leadership. Confidence level: high / medium / low on hitting number.

Deal Review Preparation

You are a sales manager preparing for a deal review with an account executive. Deal data: [PASTE: Deal name | Account | Amount | Stage | Close date | Stakeholders engaged | Last activity | Next step | Blockers identified] Build the deal review agenda: 1. Deal summary — where are we, what has happened since last review 2. Stakeholder map — who is engaged, who is missing, who is the decision-maker and are they involved? 3. Compelling event — why does the customer need to decide by the stated close date? 4. Blockers — what is preventing this deal from advancing? What is the plan to remove each? 5. Next 2 actions — specific, agreed actions with deadlines that will advance this deal Tone: Coaching, not interrogating. The goal is to help the rep, not catch them out. Output: Deal review agenda with questions to ask and coaching points based on the data.

Pipeline Generation Gap Analysis

You are a revenue operations manager identifying pipeline generation gaps. Data: [PASTE: Rep | Quota | Current pipeline | Pipeline coverage ratio | New pipeline added this month | Average deal size | Win rate % | Sales cycle length (days)] For each rep: 1. Required pipeline = Quota ÷ Win rate — how much pipeline is needed to hit quota? 2. Coverage gap — current pipeline vs. required; gap in $ 3. Pipeline generation rate — new pipeline added this month; is it sufficient to maintain required coverage? 4. Burn rate — pipeline being closed (won + lost) faster than it's being added? 5. Recommendation: pipeline generation coaching / deal quality review / quota adjustment discussion Output: Pipeline gap analysis by rep. Total team pipeline vs. required. Reps requiring pipeline generation coaching vs. those with pipeline but low conversion. Action plan.

Multi-Threaded Deal Assessment

You are a sales manager reviewing deal risk related to stakeholder engagement. Deal data: [PASTE: Deal name | Account | Amount | Stakeholders engaged (name and title) | Last contact date per stakeholder | Champion strength (strong/neutral/weak) | Economic buyer status (engaged/not engaged/unknown)] For each deal: 1. Single-threaded risk — deals where only one contact is engaged; if that person leaves or goes cold, deal is at risk 2. Economic buyer gap — deals where the economic buyer is not engaged; these rarely close 3. Champion strength — weak champion = deal is at risk even if economic buyer is engaged 4. Stakeholder map completeness — are all key buying roles identified (technical buyer / champion / economic buyer / end users)? 5. Recommended action per deal: expand contacts / re-engage cold stakeholder / escalate executive sponsor Output: Deal stakeholder risk assessment. Single-threaded deals highlighted. Economic buyer gap list. Actions to reduce deal risk through better multi-threading.

Late-Stage Deal Risk Review

You are a sales director reviewing late-stage deals for forecast risk. Late-stage pipeline: [PASTE: Deal | Stage | Amount | Close date | Days in current stage | Last customer activity | Outstanding legal/procurement steps | Any known competitors] For each deal: 1. Procurement/legal risk — is there a contract, legal, or procurement process that could delay close? 2. Budget risk — has budget been confirmed and approved? Or is it verbal only? 3. Timing risk — does the customer have a real deadline or is the close date wishful thinking? 4. Competitive risk — is a competitor still actively engaged at this stage? 5. Overall risk classification: low / medium / high — and the single most important action to derisk Output: Late-stage deal risk register. High-risk deals with specific derisking actions and owner. Forecast adjustment recommendations.

Deal Velocity Report

You are a sales operations analyst tracking deal velocity trends. Data (last 12 months of closed deals): [PASTE: Deal | Won/Lost | Amount | Stage 1 entry date | Close date | Total days to close | Number of activities | Number of stakeholders engaged] Analyze: 1. Average sales cycle length by deal size tier (small/mid/large) 2. Velocity trend — are deals closing faster or slower than 6 months ago? 3. Activity correlation — do deals with more activities close faster or slower? 4. Won vs. lost velocity — do lost deals drag on longer than won deals? 5. Fastest-closing deals — what do our fastest-closing won deals have in common? Output: Deal velocity analysis. Cycle time by deal size. Won vs. lost comparison. Top 3 factors that correlate with faster close. Recommendations.

Competitive Deal Intelligence Brief

You are a sales manager preparing a competitive intelligence brief for a deal. Deal context: [DESCRIBE: Customer, deal size, stage, our solution being proposed, known competitors in the deal, any competitive information gathered from the customer] Build the competitive brief: 1. Competitor overview — strengths and weaknesses relevant to this specific customer's needs 2. Where they will attack us — likely objections or FUD the competitor will raise about our solution 3. Where we win — our genuine differentiated strengths for this customer's use case 4. Traps to set — questions to ask the customer that highlight competitor weaknesses without naming the competitor 5. Landmines to defuse — customer concerns about our solution that need to be addressed proactively Output: Competitive deal brief. Talking points for next customer conversation. Questions to ask. Objections to prepare for.

Deal Desk Request Review

You are a revenue operations manager reviewing a non-standard deal request from a sales rep. Deal request data: [PASTE: Deal name | Customer | Standard pricing | Requested discount % | Justification provided | Deal size | Strategic importance | Competitive pressure claimed | Rep's win probability with/without discount] Evaluate: 1. Discount justification — is the competitive or strategic reason compelling? 2. Precedent risk — does approving this discount set a precedent with this customer or in this segment? 3. Margin impact — deal value at requested discount vs. standard; gross margin impact 4. Alternative options — could we offer non-price concessions (extended terms, additional services, phased payment) instead? 5. Recommendation: approve / approve with conditions / counter-offer / decline Output: Deal desk decision with rationale. Any conditions attached to approval. Counter-proposal if not approving as requested.

Pipeline Hygiene Audit

You are a sales operations manager running a pipeline hygiene audit. Pipeline data: [PASTE: Deal name | Owner | Stage | Amount | Create date | Close date | Last activity date | Last stage change date] Flag deals requiring cleanup: 1. No activity in >21 days — stalled; require rep to update or mark as lost 2. Close date in the past — overdue; require updated close date or close as lost 3. In early stage for >90 days — either advance or disqualify 4. Amount of $0 or blank — incomplete record 5. No next step recorded — requires rep to define and log next action Output: Hygiene audit report — total records reviewed, issues by type, records requiring action. Flag list to assign to reps with a 5-business-day deadline to clean up or close.

Prompts prepare work, they do not change records

Use approved CRM information, and keep record changes, configuration and customer contact with the accountable owner.

→

12 Claude-Ready CRM Skills

Downloadable Claude Skill packages for repeatable CRM work. Each defines inputs, output, limits and a reviewer.

Pipeline health review

# Pipeline health review ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a pipeline health review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a pipeline health review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Deal review packet

# Deal review packet ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a deal review packet from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a deal review packet from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Account plan summary

# Account plan summary ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare an account plan summary from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare an account plan summary from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Data quality exception review

# Data quality exception review ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a data quality exception review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a data quality exception review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Lead routing review

# Lead routing review ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a lead routing review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a lead routing review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Handoff checklist

# Handoff checklist ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a handoff checklist from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a handoff checklist from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Win/loss debrief

# Win/loss debrief ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a win/loss debrief from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a win/loss debrief from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Reporting requirements spec

# Reporting requirements spec ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a reporting requirements spec from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a reporting requirements spec from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

CRM configuration change review

# CRM configuration change review ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a crm configuration change review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a crm configuration change review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Adoption gap review

# Adoption gap review ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare an adoption gap review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare an adoption gap review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Platform evaluation packet

# Platform evaluation packet ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a platform evaluation packet from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a platform evaluation packet from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

Migration readiness review

# Migration readiness review ## Role You are a CRM workflow assistant. Prepare evidence-based reviews and drafts; do not change records, alter configuration, or contact customers. ## Required input - approved CRM exports, definitions and process documentation with the source and date ## Your task Prepare a migration readiness review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. ## Working method 1. Confirm the organization, time period, owner, and source material. List missing inputs before you begin. 2. Use only the supplied or approved connected sources. Never invent facts, policy, or context. 3. Prepare a migration readiness review from approved CRM information. Keep record changes, configuration changes and customer contact with the accountable owner. 4. Separate confirmed facts from assumptions or hypotheses. Cite source names, report lines, or record IDs when available. 5. Flag exceptions by urgency, risk, and accountable owner. ## Return this format 1. Executive summary (3-5 bullets) 2. Detailed evidence and analysis 3. Exceptions, open questions, and required evidence 4. Owner, priority, and next step for each unresolved item 5. Review checkpoint: a review-ready packet with gaps, sources and next steps ## Human review boundary the record owner or administrator approves every change ## Non-negotiable guardrails - Do not change source records, send external communications, make commitments, or make a legal, financial, or policy decision. Ask for human review whenever the evidence is incomplete or the policy is unclear.

+Suggest an edit

Connect Your Work

Start with approved exports, then connect a system once an administrator has reviewed access. Worth knowing before you compare: a vendor’s embedded assistant and an outside assistant reaching your data are different things, and the platforms differ more on the second than their marketing suggests.

Easy start

Start with a limited, low-risk connection your team can test quickly.

Approved Exports & Documentation

Access: approved CRM exports, field definitions, process docs and reports
Useful for: data-quality reviews, reporting specs and platform comparison with no integration at all
Setup & limit: work from controlled copies with a named reviewer; strip customer contact details you do not need for the task

Google Workspace

Access: Gmail, Drive and Calendar read and write for the signed-in user
Useful for: pulling correspondence and documents into an account review
Setup & limit: attachment content is not read and images inside documents are skipped; on a managed domain an administrator enables it first

Zoho CRM

Access: four separate servers covering insights, data operations, customization and workflow
Useful for: querying records without granting the ability to change them
Setup & limit: read and write live on different servers, so add only what the job needs; setup is documented for desktop clients, and plan or region requirements are not stated

Pipedrive

Access: read and write across deals, people, organisations, activities, notes and leads
Useful for: pipeline review and record preparation
Setup & limit: permissions inherit from the user role; usage is token-metered so heavy use costs beyond the seat, and the setup flow still carries a Beta label

monday CRM

Access: read and write on boards, items and columns
Useful for: pipeline and task review inside a board-based CRM
Setup & limit: an administrator can switch external AI off entirely, allow one assistant and not another, or limit access to named workspaces; tooling is board-shaped rather than deal-shaped

Attio

Access: read, write, search, semantic search over notes and calls, reporting and read-only queries
Useful for: research and analysis across a whole workspace
Setup & limit: every workspace member can connect and no administrator kill-switch is documented, so decide your own policy before rollout; query access is limited to certain plans and analytical sweeps throttle

Access: read, write and delete, each granted as a separate scope
Useful for: call and email workflow review
Setup & limit: grant the destructive scope deliberately rather than by default; per-organisation controls over who may connect are not documented

Data Enrichment Providers

Access: search and enrich company and contact records against a licensed database
Useful for: filling gaps a CRM cannot fill from its own records
Setup & limit: enrichment consumes credits and an assistant will iterate faster than a credit budget expects; check whether search results include contact details or whether each one costs an enrichment call

Needs an administrator

These can be useful, but someone needs to set access and permissions first.

Microsoft 365 & Teams

Access: SharePoint, OneDrive, Outlook and Teams content the user can already open; Teams is read-only
Useful for: meeting records and document history behind an account
Setup & limit: a tenant administrator must grant consent before anyone connects, and every call is logged to your Microsoft audit log

Slack

Access: search across channels, direct messages and shared files
Useful for: pulling deal and account context out of internal conversation
Setup & limit: Beta - confirm before you rely on it; an Owner must enable it

HubSpot

Access: read and write across contacts, companies, deals, tickets and activities
Useful for: research and preparing record updates for review
Setup & limit: bulk writes cap at 10 records at a time and custom validation rules including pipeline stage validations are not applied to records written this way

Microsoft Dynamics 365

Access: create, read, update and delete on permitted platform tables
Useful for: a governed reporting or record-review workflow
Setup & limit: an administrator configures the environment and approves which clients may connect; usage is metered in credits, and the sales-specific server does not support every client

Conversation Intelligence

Access: read-only summaries and insight from recorded calls
Useful for: adding call context to a deal or account review
Setup & limit: platforms differ sharply - some return only synthesised insight rather than transcripts, some meter each question against credits, and cadence data is often not queryable

Integration project

Plan the use case, source data, permissions, and owner before connecting.

Salesforce

Access: read, and optionally write, on permitted objects, honouring field-level security and sharing rules
Useful for: one governed reporting or research workflow
Setup & limit: there is no click-to-connect connector; an administrator must create and authorise an application, and the ChatGPT path is Beta with paid add-ons on Enterprise editions

Your Data Warehouse

Access: governed, role-based read against curated models, with optional write
Useful for: revenue reporting answered from a governed source rather than a CRM export
Setup & limit: the connection is the easy part and the data model is the project; keep any raw query tool on a separate server from a governed one, or it bypasses your semantic layer

Your Internal Systems via MCP

Access: exactly the tools you choose to expose, bound by scope and your own permission model
Useful for: one narrow, review-only recurring job before anything writes
Setup & limit: the protocol cannot enforce approval on its own, so the controls are which tools exist and how narrowly the token is scoped; ship read and search first and gate every write behind a named approver

Good to know

Useful limits and honest gaps to keep in mind before you connect anything.

What to Avoid at First

Access: do not begin with broad write access, full customer contact records, or anything that can send on a person’s behalf
Useful for: choosing a small, reversible first connection
Setup & limit: never let AI create, update or delete records unattended, change configuration or permissions, send a customer message, or approve a quote; a bad write across thousands of records is far harder to unwind than a bad read

No official path

CRMs Without an Outside AI Path

Access: none for CRM records on Freshsales, SugarCRM or Creatio
Useful for: setting expectations before an evaluation goes too far
Setup & limit: each ships a capable embedded assistant, which is a different thing; work from approved exports, and treat anything advertised as a connector for these as unofficial unless the vendor documents it

Governance separates these platforms more than features do

Ask who can enable outside AI access, whether it can be scoped to a team, and whether delete is a separate grant. Vendors rarely lead with those answers.

→

+Suggest an edit

CRM Platforms: What to Evaluate

A buyer’s starting point, not a ranking. Confirm edition, region, licensing and AI availability with the vendor before you rely on anything here - this area changes month to month.

Salesforce

Broad enterprise platform with a large partner ecosystem and its own AI portfolio.
Outside AI access: no first-party Claude connector; an administrator configures an application against the hosted servers, and the ChatGPT path is Beta with paid add-ons on Enterprise editions.
Governance: access honours field-level security and sharing rules, so an assistant sees exactly what the signed-in user sees.

HubSpot

All-in-one platform unifying marketing, sales and service records.
Outside AI access: read and write across most CRM objects, enabled by an administrator.
Governance: bulk writes cap at 10 records at a time, and custom validation rules including pipeline stage validations are not applied to records written this way - route changes through a reviewer.

Microsoft Dynamics 365

Fits organisations already standardised on Microsoft 365.
Outside AI access: table-level create, read, update and delete through the platform data server; the sales-specific server does not support every client.
Governance: an administrator configures the environment and approves clients before anyone connects, and usage is metered in credits.

Zoho CRM

Broad product suite at a lower price point than the enterprise platforms.
Outside AI access: official, split across four separate servers - insights, data operations, customization and workflow.
Governance: read and write live on different servers, so granting query access does not grant the ability to change records. Setup is documented for desktop clients.

Freshsales

Part of the Freshworks suite, with its own embedded assistant.
Outside AI access: none. Freshworks runs an official server, but it covers the service products, not Freshsales.
Governance: not applicable. Two things look like a connector and are not - the developer build tool, and a marketplace app that lets the vendor assistant act on data rather than letting yours read it.

Pipedrive

Activity-based CRM built for inside sales teams.
Outside AI access: read and write across deals, people, organisations, activities, notes and leads, on every plan.
Governance: permissions inherit from the user role. Usage is token-metered, so heavy use costs beyond the seat, and the setup flow still carries a Beta label.

monday CRM

Board-based CRM from a visual work-management platform.
Outside AI access: read and write on boards, items and columns, on all plans.
Governance: the strongest controls in this set - a master switch for external AI, per-assistant toggles, and the option to limit access to named workspaces. Because it is board-based, tooling is generic rather than deal-shaped.

Copper

Built around Google Workspace for relationship-led teams.
Outside AI access: no documented first-party path for an outside assistant; plan on working from approved exports.
Governance: not applicable until a path exists. Its Workspace integration is the practical route to context.

Creatio

No-code platform for teams that want to shape their own workflows.
Outside AI access: none for CRM records. Its one official server is a developer toolkit for building applications on the platform.
Governance: not applicable. The vendor’s own agents call out to other systems; nothing documented calls in.

SugarCRM

Flexible deployment options including on-premises.
Outside AI access: none documented.
Governance: not applicable. A prominent marketplace integration puts a chat assistant inside the product rather than letting an outside assistant reach the data.

Attio

Newer platform built on a flexible data model.
Outside AI access: the broadest reviewed - read, write, search, semantic search over notes and calls, reporting, and read-only queries.
Governance: the weakest of the capable platforms. Every workspace member can connect and no administrator kill-switch is documented - workable for a ten-person team, a real gap at a hundred. Query access is limited to certain plans.

Built for inside sales teams working the phone and inbox.
Outside AI access: read, write and delete, each granted separately.
Governance: the clearest scope model here - delete is its own grant, so destructive capability is a deliberate decision rather than a side effect.

Test the outside path, not the demo

Nearly every vendor ships an embedded assistant. Far fewer let your own assistant reach the data. Those are different purchases, and only one of them survives a change of AI tool.

→

+Suggest an edit

Tools

Products a CRM team may add around the platform. Evaluate each through your own security, privacy and procurement process - a tool is not automatically a permitted connection.

12 AI Agents for CRM Teams

An agent prepares one recurring job from approved records for a named reviewer. It never changes a record or contacts a customer.

What makes an AI agent useful

A good first agent handles one task your team already does. Give it the information it needs, tell it when to stop, and have a person check the work.

Pipeline hygiene reviewer

# Pipeline hygiene reviewer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Data quality exception monitor

# Data quality exception monitor

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Handoff gap organizer

# Handoff gap organizer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Configuration change reviewer

# Configuration change reviewer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Adoption digest

# Adoption digest

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Duplicate record reviewer

# Duplicate record reviewer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Stalled deal organizer

# Stalled deal organizer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Escalation pattern organizer

# Escalation pattern organizer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Report definition reviewer

# Report definition reviewer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Approval exception organizer

# Approval exception organizer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Partner pipeline organizer

# Partner pipeline organizer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

Adoption and benefit summarizer

# Adoption and benefit summarizer

## Job to be done
Prepare recurring accounting work for qualified human review.

## When it runs
On an approved schedule or trigger.

## Approved context
Only the approved systems, files, and records named by the workflow owner.

## Operating loop
Gather context, apply the approved skill, prepare a work packet, and route exceptions.

## What it delivers
A review-ready work packet with evidence, questions, and next steps.

## Human owner
A named accounting owner reviews and approves the output.

## What it must never do
Never take a consequential action without explicit human approval.

## How to measure it
Measure accuracy, cycle time, exception resolution, and human override rate.

## Operating rules
1. Use only approved sources and follow the linked accounting skill.
2. Cite source records and separate facts from hypotheses.
3. Route missing evidence, threshold breaches, and material exceptions to the human owner.
4. Keep an auditable record of trigger, inputs, output, reviewer, decision, and override.
5. Stop and ask for human review when the policy, evidence, or approval authority is unclear.

+Suggest an edit

Lead Scoring & AI Routing

Review how leads are scored and routed, using your own historical outcomes. The routing rules and the qualification decision stay with your team.

Predictive Scoring Models

Compare scored leads against recorded outcomes: mL trained on historical won/lost deals
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

6senseBomboraMadKuduClearbitSalesforce EinsteinHubSpot Lead ScoringDemandbaseLeadspaceG2 Buyer IntentLeadfeeder

Intent Data Integration

Compare scored leads against recorded outcomes: bombora: company-level topic surge signals
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

AI-Powered Routing

Compare scored leads against recorded outcomes: round-robin routing with performance-weighted assignment
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Qualification Automation

Compare scored leads against recorded outcomes: aI SDR handles initial qualification conversations
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Score Monitoring

Compare scored leads against recorded outcomes: score distribution dashboards for lead quality visibility
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Why It Matters

Compare scored leads against recorded outcomes: prioritization determines where reps spend time, AI scoring makes that systematic
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Lead Scoring & AI Routing Implementation Checklist

Implementation

Before You Begin

Won/lost deal data (12+ months) is clean and consistently tagged
Lead source tracking is implemented consistently across all channels
Ideal Customer Profile (ICP) is documented with measurable criteria
Marketing and Sales are aligned on the MQL definition and score threshold
Intent data vendor is contracted or actively trialed

After You're Live

Lead scores visible in rep daily activity views and used for prioritization
High-score leads contacted within 1 business day SLA consistently
Score accuracy reviewed monthly against actual conversion rates
Routing rules tested and producing balanced, fair rep assignment
Nurture sequences live and active for below-threshold leads

AI Guardrails

AI scoring models can reflect historical bias, audit for demographic or segment skew quarterly
Models need retraining when ICP shifts or product lines change significantly
Score threshold for MQL requires explicit marketing and sales alignment before launch
Intent data providers have different methodologies, validate overlap before paying for multiple
Manual overrides must be logged to provide feedback signal to improve model over time
Avoid full automation on strategic or enterprise accounts, human judgment required
Review model performance every quarter and retrain if conversion correlation degrades

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Sales Forecasting

Assemble the evidence behind a forecast number so a manager can test each assumption. People still call the number.

Signal-Based Forecasting

Assemble forecast evidence with each assumption labelled: aggregates signals from email engagement, meeting frequency, CRM stage, and call data
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

ClariAvisoGong ForecastSalesforce Forecast AIPeople.aiBoostup.aiInsightSquaredChorus.aiHubSpot ForecastingScratchpad

Pipeline Inspection AI

Assemble forecast evidence with each assumption labelled: deal health scores for every open opportunity in real time
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Conversation Intelligence Signals

Assemble forecast evidence with each assumption labelled: gong and Chorus extract deal signals from call transcripts
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Manager Tools

Assemble forecast evidence with each assumption labelled: aI-generated forecast submission summaries per rep
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Integration Ecosystem

Assemble forecast evidence with each assumption labelled: cRM (Salesforce, Dynamics, HubSpot) as primary data source
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Why It Matters

Assemble forecast evidence with each assumption labelled: traditional forecasting relies on rep self-reporting, which is optimistic, inconsistent, and late
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Sales Forecasting Implementation Checklist

Implementation

Before You Begin

CRM deal data is current, updated within 2 weeks per opportunity
Sales stages have clear, objective exit criteria defined and documented
Email and calendar are connected to CRM for automatic activity capture
Conversation intelligence tool is recording and transcribing all sales calls
Historical actuals (8+ quarters) are in the system for model training

After You're Live

AI forecast reviewed alongside rep commits in every weekly forecast call
Pipeline health dashboard is the manager's primary inspection tool
At least one risk flag addressed per rep per week in coaching
Forecast accuracy tracked against actuals every month
AI forecast call is within 10% of actual closed revenue at quarter end

AI Guardrails

AI forecasting is only as good as CRM data quality, enforce update discipline as a team norm
Use AI as input to the forecast call, not a replacement for human judgment
Beware sandbagging: AI models trained on artificially low commits will reflect that bias
Deal exclusions must be documented to avoid distorting historical accuracy metrics
Forecasting tools need 2+ quarters of consistent data before baselines are reliable
Present AI signals in context, not as final verdicts that override rep knowledge
Audit for systematic bias in opportunity scoring across segments and reps

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Conversation Intelligence

Summarise recorded calls your team is permitted to use, traceable back to the call. Coaching judgement stays with the manager.

Auto-Recording & Transcription

Summarise permitted call records: automatic meeting and call recording with consent disclosure
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

GongChorus.aiWingman (Clari)Salesloft ConversationsOutreach KaiaJiminnyFireflies.aiAvomatl;dvOtter.ai

Deal Signal Extraction

Summarise permitted call records: budget, authority, timeline, and need mentions tracked per call
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Coaching & Playbooks

Summarise permitted call records: talk-to-listen ratio tracked per rep vs. top performer benchmark
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Win/Loss Pattern Analysis

Summarise permitted call records: pattern analysis across won vs. lost deal call libraries
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Team Intelligence

Summarise permitted call records: cross-rep trend analysis for playbook refinement
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Integration Ecosystem

Summarise permitted call records: cRM sync pushes summaries to Salesforce, HubSpot, or Dynamics
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Conversation Intelligence Implementation Checklist

Implementation

Before You Begin

Recording consent language is included in all meeting invitations
Call recording policy is communicated clearly to the entire sales team
CRM integration is connected, mapped, and tested for automatic sync
Managers have committed to a coaching cadence using recordings
Deal signal keywords are configured to match your specific sales motion

After You're Live

80% or more of sales calls are recorded and transcribed consistently
Call summaries pushed to CRM opportunity records automatically after every call
Managers review at least 2 calls per rep per month in coaching sessions
Competitive mentions tracked and reported to product and marketing monthly
Win/loss patterns reviewed quarterly for playbook updates and training

AI Guardrails

Two-party call recording consent is legally required in many US states and international markets, implement disclosure statements in all meeting invites
Recording sensitive enterprise negotiations requires legal review before sharing recordings
AI transcription accuracy varies by accent and background noise, spot-check critical calls
Never use call scoring as the sole performance metric, context and deal specifics matter
Configure competitor keyword tracking carefully, false positives create review fatigue
Ensure recording storage meets data retention and deletion policies

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Email & Sequence Automation

Draft and review outreach from approved messaging. Sending stays with the person whose name is on it.

AI Email Generation

Draft from approved messaging: generate first drafts from contact, company, and deal context
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

OutreachSalesloftApollo.ioLavenderReply.ioInstantly.aiLemlistMixmaxKlentyMailshake

Sequence Automation

Draft from approved messaging: multi-step cadences combining email, call, LinkedIn, and SMS
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Deliverability AI

Draft from approved messaging: inbox placement optimization with provider-specific tuning
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Personalized Outreach

Draft from approved messaging: dynamic content blocks populated from CRM data fields
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Analytics & Optimization

Draft from approved messaging: open, click, and reply rate by template, sequence, and rep
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Compliance Automation

Draft from approved messaging: gDPR and CCPA-compliant opt-out link insertion in all emails
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Email & Sequence Automation Implementation Checklist

Implementation

Before You Begin

Email domain is authenticated (SPF, DKIM, DMARC) and domain is warmed
Unsubscribe and opt-out processes are documented and legally compliant
ICP is defined with enough specificity for AI personalization to be meaningful
Sequence stages align to your sales process stages in CRM
Contact data has sufficient enrichment context fields for AI personalization

After You're Live

AI-generated emails reviewed by rep before every single send, no fully automated sends
Sequence performance reviewed weekly with step-level data by manager
Reply rate improving month over month tracked as a team KPI
Deliverability monitored monthly (inbox placement rate, spam rate)
Sequences updated quarterly based on performance data and market changes

AI Guardrails

Never auto-send AI-generated emails without human review, personalization errors cause significant damage
CAN-SPAM and GDPR require physical address and clear opt-out in all commercial emails
Over-sequencing damages domain deliverability and brand reputation, enforce contact frequency limits
AI personalization using LinkedIn data may violate platform terms of service, use only official licensed data partnerships
Monitor reply sentiment, negative replies should immediately exit sequences and trigger rep review
Test new sequences on small cohorts before rolling out broadly

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Pipeline Management

Turn recorded deal activity into review packets and exception lists for a manager.

Deal Health Scoring

Build an exception list from recorded activity: multi-signal health score per opportunity combining all data sources
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

ClariBoostup.aiSalesforce Pipeline InspectionHubSpot Deal ManagementGong Deal BoardsPeople.aiScratchpadOutreach Deal InsightsAvisoRevenue.io

Activity Intelligence

Build an exception list from recorded activity: email, call, and meeting frequency tracked per deal automatically
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Stage Progression AI

Build an exception list from recorded activity: predictive time-to-close per deal based on current signals
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Risk & Coverage

Build an exception list from recorded activity: single-threaded deal risk flags with manager escalation
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Inspection Tools

Build an exception list from recorded activity: aI-generated pipeline summary for weekly forecast meeting prep
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Pipeline Analytics

Build an exception list from recorded activity: pipeline waterfall report showing created, moved, and closed by period
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Pipeline Management Implementation Checklist

Implementation

Before You Begin

Sales stages are defined with objective, measurable entry and exit criteria
Activity capture (email and calendar sync) is fully configured for all reps
Deal age benchmarks established from historical closed deal data
Manager pipeline review cadence is agreed and scheduled
Quota and territory data is accurate and loaded in CRM

After You're Live

Managers using AI deal health scores as primary inspection tool in pipeline reviews
At-risk deals identified and actioned within the same working week
Activity gaps resolved within 48 hours of receiving alert
Pipeline coverage ratio reviewed every week vs. quota target
At least one process improvement implemented per quarter from pipeline analytics

AI Guardrails

AI deal health only reflects data that is logged, enforce activity logging or signals are missing and scores are wrong
Deal health scores should supplement manager judgment, not replace it
Pipeline inflation (ghost deals) will distort AI models, enforce regular deal hygiene and close cycles
Don't use at-risk flags to penalize reps, use them to coach and support
Customize AI risk thresholds for different deal types and average sales cycles
Single-threaded flags need context, some deals have a single buyer by design

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Contact & Account Intelligence

Check what CRM records are missing or stale and where a licensed source could fill the gap. Writing the record stays with its owner.

Automated Enrichment

Index what is missing, stale or conflicting: real-time contact data fill on every new record creation
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

ZoomInfoClayApollo.ioClearbitBombora6senseLushaCognismHunter.ioBuiltWithDatanyzeLeadfeeder

Intent Signal Integration

Index what is missing, stale or conflicting: topic surge: which companies are actively researching your category
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Data Quality Management

Index what is missing, stale or conflicting: duplicate detection and merge in bulk with configurable rules
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Clay-Style Research AI

Index what is missing, stale or conflicting: multi-source waterfall enrichment logic with fallback providers
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Account-Based Intelligence

Index what is missing, stale or conflicting: full account map showing all known contacts per company
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Technology Intelligence

Index what is missing, stale or conflicting: builtWith and Datanyze tech stack detection per company
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Contact & Account Intelligence Implementation Checklist

Implementation

Before You Begin

Primary enrichment vendor is contracted and integrated with CRM
Existing contact and account data is audited for quality and completeness
Duplicate records are identified and merged before enrichment begins
Intent data strategy is defined (which topics, which providers, which segments)
GDPR and CCPA data handling processes are documented and reviewed by legal

After You're Live

New records enriched automatically within minutes of creation
Monthly data decay rate below 5% on key contact fields
Intent signal alerts acted on by sales team within 2 business days
Duplicate rate in CRM below 3% on ongoing basis
Technology signals incorporated into lead scoring model as a factor

AI Guardrails

Third-party enrichment data cannot be resold or shared, license restrictions apply
Email addresses from enrichment tools require explicit consent before cold outreach to EU contacts
Data freshness varies significantly by provider, tier accounts for enrichment frequency and cost
Tech stack detection is probabilistic not definitive, verify for critical enterprise accounts
Intent data is company-level not individual-level, use appropriately in outreach personalization
Never assume enriched data is fully accurate, spot-check on all strategic accounts
Comply with LinkedIn's restrictions on data scraping, use only official licensed data partnerships

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Marketing Automation

Review campaign, routing and handoff rules against what the records show happened.

Behavioral Automation

Restate documented rules and show where records diverge: trigger-based emails from website visit and content download activity
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

MarketoHubSpot Marketing HubPardotKlaviyoActiveCampaignBrazeDemandbase6senseDriftMutiny

Account-Based Marketing

Restate documented rules and show where records diverge: target account list synchronized with CRM opportunity pipeline
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

AI Content Optimization

Restate documented rules and show where records diverge: subject line and CTA optimization with statistical A/B testing
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Lead Lifecycle Management

Restate documented rules and show where records diverge: mQL definition enforced automatically with AI scoring threshold
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Multi-Channel Orchestration

Restate documented rules and show where records diverge: email, SMS, LinkedIn, and push notification automation coordinated
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Revenue Attribution

Restate documented rules and show where records diverge: first-touch, last-touch, and multi-touch attribution models compared
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Marketing Automation Implementation Checklist

Implementation

Before You Begin

Marketing and Sales are explicitly aligned on MQL, SQL, and SQO definitions
CRM-to-MAP integration is bi-directional, mapped, and tested in both directions
Email domain is authenticated and sending reputation is established
UTM tracking is implemented consistently across all campaigns
Content library is organized and tagged by buyer stage and persona

After You're Live

AI-scored leads trigger automatic enrollment in appropriate nurture tracks
Sales receives MQL alerts with full behavioral and firmographic context
Marketing pipeline contribution reported monthly in revenue review
Attribution model agreed upon between marketing and finance leadership
Content performance reviewed monthly with AI-generated optimization recommendations

AI Guardrails

Marketing automation plus AI equals scale, which magnifies both success and mistakes, so start with small cohorts
AI content generation requires brand voice guidelines and mandatory human review before publishing
Email marketing opt-in consent is required in most markets, enforce strictly and test regularly
Frequency caps prevent list fatigue, set and enforce maximum email contacts per month
Unsubscribe requests must be processed within regulatory timeframes
Attribution models have inherent bias, use multiple models and triangulate results

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Customer Service AI

Organise case and escalation records for review. Case decisions and customer replies stay with the service team.

AI-Powered Deflection

Index case evidence and escalation history: chatbot handles Tier 1 questions without agent involvement
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Zendesk AISalesforce Service Cloud EinsteinIntercom FinFreshdesk FreddyKustomerHubSpot Service HubTidioDriftGainsightChurnZero

Agent Assist

Index case evidence and escalation history: real-time response suggestions during active ticket handling
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Case Intelligence

Index case evidence and escalation history: automatic case categorization and topic tagging at intake
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Customer Health Monitoring

Index case evidence and escalation history: churn risk prediction from support frequency and sentiment patterns
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Automation & Routing

Index case evidence and escalation history: auto-assignment by agent skill, capacity, and topic match
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Knowledge Management AI

Index case evidence and escalation history: auto-generate help articles from resolved ticket content
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Customer Service AI Implementation Checklist

Implementation

Before You Begin

Knowledge base is documented, organized, and kept current by category
CRM and helpdesk are integrated with bi-directional data sync tested
Support tier structure and escalation paths are defined and documented
CSAT and NPS surveys are implemented with baseline data established
Churn risk definition agreed across CS, Sales, and Finance leadership

After You're Live

AI chatbot deflecting at least 30% of inbound tickets without human intervention
Agent assist surfacing relevant article suggestions on 70% or more of cases
Case categorization automated with 85%+ accuracy verified monthly
Health scores used in monthly renewal and expansion review meetings
At-risk accounts flagged by service AI actioned by CSM within 48 hours

AI Guardrails

AI chatbot frustrates customers if it cannot escalate smoothly, design human handoff with extreme care
Never use AI to deny claims or reject high-stakes customer requests without human review
Customer sentiment data is sensitive, implement role-based access restrictions
AI case prioritization may deprioritize legitimate issues with low volume, review edge cases monthly
GDPR subject access requests cannot be handled by AI alone, require human review
Ensure accessibility compliance for all AI-powered self-service portal experiences

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

Revenue Analytics

Prepare revenue reporting from approved definitions, with each figure traceable to its source.

Revenue Intelligence

Prepare reporting from approved definitions: deal-level signal aggregation from all CRM and activity sources
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

ClariTableauPower BISalesforce Einstein AnalyticsGong InsightsLookerChorus AnalyticsThoughtSpotDomoCoefficient

Predictive Analytics

Prepare reporting from approved definitions: revenue forecast with confidence intervals and variance tracking
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Activity Analytics

Prepare reporting from approved definitions: email, call, and meeting correlation to revenue outcomes
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Cohort & Attribution

Prepare reporting from approved definitions: customer cohort revenue analysis by acquisition period and source
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Executive Dashboards

Prepare reporting from approved definitions: board-ready revenue dashboards with AI narrative commentary
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Embedded Analytics

Prepare reporting from approved definitions: analytics embedded inside CRM without switching to separate tools
What to review: Confirm the source of each fact, what the record shows, and what is still an assumption
Control: The record owner approves any record change, configuration change, or customer communication

Revenue Analytics Implementation Checklist

Implementation

Before You Begin

Single source of truth is defined with CRM as the system of record
Revenue definitions (ARR, MRR, TCV, ACV) documented and agreed across teams
Quota and territory data is accurate and current in CRM
Historical actuals (8+ quarters) available for model training and trend analysis
BI tool is connected to CRM as the primary data source

After You're Live

AI forecasting used as primary tool in every weekly forecast meeting
Revenue dashboard shared with executive team weekly with AI commentary
Rep performance data used in structured 1:1 coaching conversations
Churn risk model reviewed in monthly CS and CS leadership meeting
At least one revenue insight has driven a process or strategy change in the last quarter

AI Guardrails

Revenue analytics are only as accurate as the underlying data, invest in data quality before investing in analytics tools
Rep-level performance data is sensitive, implement and enforce role-based access controls
AI predictions are probabilistic, present confidence ranges not single-point estimates
Attribution models are never perfect, use multiple and triangulate for strategic decisions
Historical analysis can embed past biases, review segmentation assumptions regularly
Don't optimize solely for AI-favored metrics, qualitative signals and rep knowledge matter too
Align all analytics definitions with finance before reporting to board or investors

Implementation Note

Start with one function, measure results for 60 days, then expand. Implementing too many functions at once leads to lower adoption across all of them.

💡

+Suggest an edit

CRM Buyer’s Checklist

Use these in demos, renewals and implementation planning. Ask for written answers and evidence that matches your own data, not a reference account’s.

Run a Structured Demo

Ask vendors to demo AI features on a dataset that resembles your own, not a pre-loaded demo environment with perfect data
Request a live walk-through of lead scoring logic: why did it score this specific lead this way?
Ask what happens when the AI is wrong, how does a rep override it, and does that feedback improve the model?
If a feature is “coming soon,” ask for a specific GA date in writing, not a roadmap slide

Vet the AI Claims

Ask whether each AI feature is generally available or still in beta, vendors often present roadmap as product
Ask which customers in your segment are using the AI feature in production today, not in pilot
Ask how long the AI has been in market and how the model improves over time
Ask whether AI features require a separate add-on SKU or are included in base licensing

Assess Your Data Fit

Most AI CRM features require clean, consistently tagged historical data to generate useful outputs
Ask vendors what data quality baseline their AI needs, and what happens below that threshold
Evaluate your current CRM data: are won/lost deals tagged, are lead sources tracked, are contacts enriched?
If your data is poor, prioritize platforms with strong data cleanup tooling before evaluating AI capabilities

Get the Right References

Ask for references in your exact segment: same industry, similar deal size, similar team structure
Vendor-provided references are pre-screened, ask for the G2 or Gartner Peer Insights community, not just their list
When you call references, ask specifically: which AI features are you actually using day-to-day?
Ask what they would do differently if they were buying again, that answer is more useful than any case study

Model the Full Cost

The license fee is rarely the largest cost, factor in implementation, integrations, training, and internal time
Ask what is included in the stated price vs. billed separately: storage, API calls, AI credits, additional seats
Get a total cost of ownership estimate for year one and year three, not just the annual contract number
Confirm data portability and export terms before signing, switching costs are real

Know the Red Flags

AI features that require a separate enterprise tier or add-on to access, check what you’re actually buying
No explainability on model outputs, if the AI can’t tell reps why a lead was scored a certain way, they won’t trust it
Data portability not explicitly in the contract, you should own your data and be able to export it cleanly
Vague answers on sub-processors, if they can’t tell you who handles your customer data, that’s a compliance risk

Test External AI Access

Ask whether your own AI assistant can read and write records, not just whether the platform has AI features - these are different purchases
Ask who can turn that access on, whether it can be limited to named teams or workspaces, and whether delete is granted separately from write
Run one real workflow through the external path during the trial and count what a person still had to correct

20 Questions for Your CRM Vendor

Must-Ask

AI Capabilities

1. Which AI features are generally available today vs. on the roadmap?
2. Can you show me a live customer reference using these AI features in production?
3. Does the AI work on our existing data or require migration and re-setup?
4. Can the AI explain why it scored a specific lead or flagged a specific deal?
5. What happens when the AI is wrong, how does the system handle errors and rep feedback?

Data & Integration

6. Where does my data go when the AI processes it?
7. Is my data used to train models shared with other customers?
8. Is bi-directional sync available with our email and calendar platform?
9. What is the API rate limit and how complete is the integration partner ecosystem?
10. Is data portability explicitly guaranteed in the contract terms?

Security & Compliance

11. Is SOC 2 Type II certification available and current?
12. How does the vendor handle GDPR data subject access requests?
13. Which sub-processors have access to our customer data?
14. Can I set approval thresholds for AI-automated actions?
15. What audit trail exists for AI actions and recommendations?

Cost & Contract

16. How is AI priced, per user, per feature module, or included in base?
17. What is the realistic implementation timeline and what internal resources will it require?
18. What data quality requirements must we meet before AI features work reliably?
19. What training and change management support do you provide at launch?
20. What happens to our data and AI configuration if we leave the platform?

Pro tip

If a vendor can’t clearly answer questions 1, 2, 6, and 7, they’re selling you a roadmap, not a product. Slow down before signing.

→

+Suggest an edit

Build Your Business Case

How to build a case from your own baseline. Every figure here should be yours, measured before you start - not a vendor’s.

Start With Your Baseline

Before any AI investment, document your current state: quota attainment rate, average win rate, average sales cycle length, and forecast accuracy
These are your before numbers. Without them, you can’t measure after.
Most teams skip this step and end up unable to prove impact, even when AI is clearly working
Pull 4-8 quarters of data from your CRM before making any purchasing decisions

What to Measure

Speed: Lead response time, time from opportunity created to first meeting, average days in each pipeline stage
Quality: MQL-to-SQL conversion, win rate by lead source, deal size by scoring tier
Efficiency: Activities logged per rep per week, time in CRM vs. time with customers
Accuracy: Forecast variance at month-end and quarter-end vs. actual closed

How to Model the Case

Pick one metric you believe AI will move, start narrow, not broad
Estimate a conservative improvement (e.g., 10% faster lead response) and calculate the revenue impact at your current pipeline volume
Compare that to the fully-loaded cost of the tool (license + implementation + training time)
If the conservative case doesn’t justify the investment, the aggressive case probably won’t either

What Vendors Won’t Tell You

Published benefit figures come from customer success stories, by definition the best outcomes, not the average
Most AI CRM benefits take 6-12 months to materialize as models train on your data
The biggest cost is usually not the software, it’s the internal time to clean data, train reps, and change habits
Ask vendors for customer references in your exact segment and deal size, not their showcase logos

What Actually Drives the Benefit

Time reallocation: If AI handles admin, reps can spend more time on high-value conversations, the magnitude depends entirely on your current admin burden
Better prioritization: Fewer wasted calls on the wrong leads has compounding value over a quarter
Earlier risk detection: Catching a deal slipping 3 weeks earlier than you would have is worth more than any dashboard feature
Onboarding speed: New reps ramping faster because AI surfaces the right next actions is measurable and often overlooked

What Good Looks Like

Year 1 should be about proving the concept in one function before expanding, not deploying everything at once
Pick a pilot group, set a specific goal, measure it, and let the results make the case for the next phase
If you can’t show a clear before/after on your first use case, more AI tools won’t fix the problem
The organizations that get the most from AI CRM are the ones that treat it as a process change first and a technology purchase second

Bottom line

Build your business case on your own numbers, not vendor benchmarks. Set a baseline, pick one metric to move, measure it honestly, and let results drive the next investment decision.

→

+Suggest an edit

Set Up & Get Running

Begin with one low-risk review workflow using approved exports. Connect a platform only after an administrator has reviewed who may use it and what it may change.

Implementation Timeline

Days 1–30: Foundation

Audit data quality.Deduplicate, enrich, validate required fields. Poor data defeats AI before it starts.
Map sales stages.Define objective exit criteria for each stage so AI has clean signal to work with.
Connect email and calendar.Automatic activity capture is not optional, reps won’t log every touchpoint manually.
Document ICP in writing.Measurable firmographic and behavioral criteria, debated with your top reps.
Select AI vendors.Lead scoring, conversation intelligence, enrichment. Prioritize CRM integration over feature lists.
Set baselines.Measure current conversion rates, cycle length, and forecast accuracy before any AI goes live.

Days 31–60: Activation

Launch lead scoring.Start with firmographic model (company size + industry + title). Add behavioral signals at day 60.
Deploy first email sequence.Short (3–5 steps), A/B test subject lines, measure reply rate weekly.
Roll out conversation intelligence.Get recording consent in place first. Build the habit of recording before using AI coaching.
Train the full team.Mandatory session, recorded for new hires. Measure attendance and follow up weekly.
Run AI-assisted pipeline review.Use deal health scores in manager 1:1s before the full team meeting.
Enrich all new leads on entry.Auto-enrichment on lead creation, not manually 30 days later.

Days 61–90: Optimization

Review lead score accuracy.Compare scores from Day 1 against actual outcomes. Adjust model weights by segment.
Optimize email sequences.Replace underperforming steps. Keep what’s working. Test new angles on stalled leads.
First call coaching session.Use conversation intelligence to share clips of great calls, not just mistakes.
Add AI forecasting.Run AI call alongside rep commits. Don’t replace the rep forecast yet, compare them.
Document wins.Measure productivity change, cycle length, forecast accuracy. Build the case for the next phase.
Plan next rollout.If lead scoring succeeded, add call coaching. If adoption is low, fix the model before expanding.

Implementation Success Metrics

Goals

30-Day Targets

5-10 reps trained and using AI tools daily
Baseline KPIs established (conversion rate, AHT, pipeline velocity)
AI usage guidelines documented and communicated
Daily feedback collected from pilot group

60-Day Targets

Full sales team deployed with AI tools
10-15 proven prompts in shared library
2nd AI use case integrated and live
KPI improvement measured vs. baseline

90-Day Targets

3 workflows operationalized with SOPs
AI usage policy formalized and approved
Team cross-trained (no knowledge silos)
Total benefit calculated: pipeline, conversion, time savings
Next wave planned (forecasting AI, analytics, expansion)

Sequence matters

Data quality → activation → measure → optimize. Skip steps and you’ll restart at step one.

→

+Suggest an edit

How the Pieces Fit Together

A prompt helps with one task now. A skill saves how your team does it. A connector brings approved records in. An agent runs one recurring job for a named owner.

Native CRM AI

What it means: AI built directly into the CRM by the vendor, same data model, same security, same interface
Examples: Einstein (Salesforce), Breeze (HubSpot), Zia (Zoho), Copilot (Dynamics), Freddy (Freshsales)
Upside: No integration work. AI sees the same data as users. Updates ship with the platform.
Downside: Locked into vendor AI roadmap and pricing. Limited customization.

Conversation Intelligence

What it means: Third-party tools that record, transcribe, and analyze every sales call
Examples: Gong, Chorus, Wingman, Jiminny, sit on top of any CRM via API
Upside: Surfaces coaching moments, deal risks, and competitive mentions from calls
Downside: Separate vendor and contract. Requires consistent recording adoption.

Revenue Intelligence

What it means: Platforms that aggregate signals from email, calendar, CRM, and calls to surface pipeline risk
Examples: Clari, Aviso, People.ai, works across any CRM, not locked to one platform
Upside: Earlier deal risk detection and more reliable forecasts than manual reviews
Downside: Another integration layer. Requires clean email and calendar data to work.

Data Enrichment

What it means: Tools that keep contact and account data fresh, complete, and accurate automatically
Examples: Clay, ZoomInfo, Apollo, Clearbit, typically run as a background layer enriching CRM records
Upside: Dead data defeats AI. The enrichment layer is the foundation everything else builds on.
Downside: Ongoing cost. Data quality varies by provider. Needs rules on when to auto-update vs. review.

AI Outreach & Sequences

What it means: Platforms that personalize outreach across large lists, optimize send timing, and A/B test messaging
Examples: Outreach, Salesloft, Apollo AI, Lavender, sit on top of CRM and drive pipeline activity
Upside: Increases outreach volume and message quality without proportional rep time
Downside: Over-automation risk. Poorly tuned sequences damage sender reputation.

Agentic CRM

What it means: AI that doesn’t just answer questions, it takes actions across the customer lifecycle
Examples: Salesforce Agentforce, HubSpot AI Agents, handle qualification, follow-up, scheduling, and renewals autonomously
Key question: What can the agent do without human approval? Every vendor draws this line differently.
Still early. Define your human-in-the-loop thresholds before deploying agents in customer-facing workflows.

Architecture rule of thumb

Start with native AI for core CRM processes. Layer in best-of-breed for specialized functions. Build custom only when nothing else fits.

→

+Suggest an edit

Where AI Can Help a CRM Team

AI can prepare, review and explain CRM work. It does not change records, alter configuration, or talk to customers on your behalf.

Review & Preparation

Turn approved exports into pipeline, account and data-quality reviews
Show the source and date behind every figure
The record owner decides what changes

Evaluation & Selection

Compare platforms against your own data and process, not a demo dataset
Prepare demo scripts, reference questions and cost models
Your team judges fit and signs the contract

Drafting & Documentation

Draft reporting narratives, process docs and handoff checklists
Separate what a record shows from what is still an assumption
A reviewer approves anything a customer will see

What Stays With People

Creating, updating or deleting records
Configuration, permissions and routing rules
Customer contact, quote approval and the purchase decision

A CRM is a system of record before it is an AI product

Judge a platform on what your team can trust it to hold, then on what an assistant can do with it.

→

+Suggest an edit

Controls, Data & Guardrails

CRM AI touches customer records, permissions and your system of record. Set these before the first pilot.

Data Governance

CRM is the single system of record for all customer data
Field-level ownership assigned per data domain
Data quality SLAs (freshness, completeness by field)
Automated deduplication rules running continuously
Data retention and right-to-deletion policies configured

AI Model Oversight

AI model performance reviewed against actuals every quarter
Training data sources documented for transparency
Demographic and segment bias audited in lead scoring models
Models retrained when ICP or product changes significantly
Human review threshold defined for high-stakes AI decisions

Access Controls

Role-based access to CRM data and AI features configured
Manager access to rep call recordings and performance data governed
Customer data access restricted by territory and account assignment
Admin audit log enabled for all AI configuration changes
Integration permissions reviewed and reauthorized annually

Vendor Risk Management

CRM vendor SOC 2 Type II certification verified and current
Data processing agreements (DPA) signed with all AI vendors
Sub-processor list reviewed for all tools with customer data access
Data residency requirements confirmed for regulated industries
Contract exit provisions include data export and portability guarantees

Team Policy

Acceptable use policy for AI email generation documented and signed
Call recording consent and disclosure process enforced
AI tool usage in customer-facing communications reviewed by manager
Data entry standards and CRM update frequency requirements defined
Training completion required before gaining access to AI features

Compliance

GDPR and CCPA compliance for all contact and customer data
CAN-SPAM and CASL compliance for all email automation
TCPA compliance for SMS sequences with documented opt-in
Industry-specific requirements met (HIPAA, FINRA, SOC 2 as applicable)
Regular compliance audit on quarterly schedule with documented outcomes

Incident Response

AI output error escalation process defined and tested
Data breach response plan with notification timelines documented
Customer complaint handling process for AI interaction issues
Model failure detection with rollback procedure
Communication protocol for AI incidents to team and customers

Change Management

Executive sponsor named for AI CRM program
Change management plan executed before each major rollout
Rep feedback loop with monthly structured input session
Success metrics and review cadence established at launch
Vendor QBR scheduled quarterly with AI roadmap review

Governance Checklist

Policy

Checklist

AI usage policy is documented and signed by all sales team members
Call recording consent disclosures are included in all meeting invites and sales sequences
Data processing agreements are signed with all AI CRM vendors and sub-processors
Role-based access control is configured and access lists reviewed quarterly
AI model performance is reviewed and documented every quarter against actuals
GDPR and CCPA opt-out processes are tested and confirmed compliant
Data retention and deletion policies are configured in CRM and all connected tools
Executive sponsor reviews AI CRM program metrics and roadmap monthly

The record owner owns the change

AI can prepare a record update, a routing rule or a report. A named person applies it.

→

Sections

Start Here Prompts Skills Connect Your Work Platforms Tools Agents

Set Up & Get Running

Start With Your Role

100 AI Prompts for CRM Teams

12 Claude-Ready CRM Skills

Connect Your Work

Easy start

Needs an administrator

Integration project

Good to know

No official path

CRM Platforms: What to Evaluate

Tools

AI CRM Platforms 12

Revenue Intelligence & Forecasting 10

Conversation Intelligence 10

Email & Sequence Automation 10

Data Enrichment & Intent 12

Lead Scoring & ABM 10

Marketing Automation 10

Sales Enablement 10

Customer Success & Health 10

CPQ & Contract AI 10

Data Analytics & BI 10

Communication & Scheduling 8

AI Assistants & LLMs 8

Automation & Integration 8

12 AI Agents for CRM Teams

Lead Scoring & AI Routing

Lead Scoring & AI Routing Implementation Checklist

Before You Begin

After You're Live

Sales Forecasting

Sales Forecasting Implementation Checklist

Before You Begin

After You're Live

Conversation Intelligence

Conversation Intelligence Implementation Checklist

Before You Begin

After You're Live

Email & Sequence Automation

Email & Sequence Automation Implementation Checklist

Before You Begin

After You're Live

Pipeline Management

Pipeline Management Implementation Checklist

Before You Begin

After You're Live

Contact & Account Intelligence

Contact & Account Intelligence Implementation Checklist

Before You Begin

After You're Live

Marketing Automation

Marketing Automation Implementation Checklist

Before You Begin

After You're Live

Customer Service AI

Customer Service AI Implementation Checklist

Before You Begin

After You're Live

Revenue Analytics

Revenue Analytics Implementation Checklist

Before You Begin

After You're Live

CRM Buyer’s Checklist

20 Questions for Your CRM Vendor

AI Capabilities

Data & Integration

Security & Compliance

Cost & Contract

Build Your Business Case

Set Up & Get Running

Days 1–30: Foundation

Days 31–60: Activation

Days 61–90: Optimization

Implementation Success Metrics

30-Day Targets

60-Day Targets

90-Day Targets

How the Pieces Fit Together

Where AI Can Help a CRM Team

Controls, Data & Guardrails