I connected Affinity to Claude via MCP and spent a week seeing what’s possible. Here are eight workflows that legitimately changed how I think about what a CRM can do.
Most of what dealmakers call “CRM work” is actually data retrieval like pulling pipeline reports, prepping for meetings, and checking who knows whom, and updating deal stages after a call. MCP (Model Context Protocol) lets AI assistants like Claude, ChatGPT, Gemini, and Copilot query that data directly and write back to it. Ask "Who at our firm has the strongest relationship with [company]'s CEO?" and get an answer. Then say "Update the deal stage on [company] to Due Diligence," "Add ‘Healthcare’ as a new industry tag [company]," or “Set a reminder to follow up with [company]'s CEO next Tuesday,” and it happens.
You can get deal history, relationship scores, pipeline status, and activity logs without exports, API wrappers, or engineering work. Affinity’s hosted MCP Server now makes this available for Affinity customers. Your CRM becomes the data layer that powers AI workflows across your entire firm, from automated meeting prep and pipeline analysis to relationship tracking and deal management.
Every workflow below was done by typing a natural-language prompt.
1. Sell-side bank performance analysis
An afternoon in Excel, done conversationally in minutes.
Prompt: “Pull all deals from our PE pipeline with a sell-side bank assigned. For each bank, calculate deal count and average LTM EBITDA. Chart it.”
[screenshot]
I followed up: “For banks with more than five deals, show percentage that advanced past Deeper Diligence.” Interactive donut charts by bank, with the AI explaining its own filter logic. Two prompts, two visualizations, about 25 seconds of actual work.
2. The 30-second meeting brief
You have a call in 10 minutes and haven’t touched the CRM since the last meeting. Normally, that means 15 minutes of frantic toggling between tabs.
Prompt: “Prep me for my meeting with [Company]. Deal history, strongest relationships on our team, what’s happened since our last touchpoint.”
Deal stage, team connections ranked by relationship strength, a timeline of every interaction. All in one response.
3. Pipeline health check for IC
Your IC meeting is tomorrow. The analyst who usually pulls the pipeline report is on vacation. Now what?
Prompt: "Show me all deals without activity in 30 days, grouped by stage."
The MCP pulls every deal entry along with last-contact timestamps and stage from your pipeline list. Claude filters for anything stale and groups the results. No spreadsheet needed, and no manual sorting. Because the context carries forward, each follow-up builds on what's already loaded:
"Which of these have an upcoming close date?""Which deals moved backward this quarter?""Who on our team has the strongest relationship with the company in IOI that's gone cold?"
You're interrogating your pipeline the way you'd interrogate a colleague, asking one question at a time, going deeper with each answer.
One thing to know: activity data is firm-wide. The MCP can tell you when anyone at your firm last contacted a company, but it can't filter by a specific person's sent emails. "Has the deal lead personally followed up" is a judgment call you still own.
4. Pass quality audit
Most firms never look back at their passes. This workflow changes that.
Prompt: "For every deal we passed on in the last two years, show our current relationship strength with the founding team. Flag anyone we've gone cold with where the pass was marked favorable."
In this case, Affinity has already logged the relationship strength scores, pass reasons, and interaction history. Claude just surfaces what's buried in the data. A favorable pass that's gone cold is a re-engagement opportunity your team would never find manually. Follow up with:
"Which partner led the most favorable passes that have since gone cold?"
That's better than a report. It’s a feedback loop that turns two years of pass decisions into a sharper sourcing strategy starting today.
5. Relationship decay detection
The relationships you can least afford to lose are the ones most likely to go unmonitored.
Prompt: "Which portfolio companies hasn't our team had any contact with in over 90 days? Rank by ownership stake."
Affinity has been logging every email, meeting, and touchpoint across your entire team automatically. The MCP surfaces what's buried in that data without manual tracking or status update meetings. Claude pulls last-contact timestamps across your full portfolio list, flags anything that's gone cold, and ranks by ownership stake so you're looking at your highest-exposure relationships first.
Follow up with:
"For each of those, who on our team has the strongest existing relationship?"
Now you have a list of at-risk relationships with a name next to each one. The person most likely to get a reply is already identified. That's the difference between a report and an action plan.
6. Deal research agent
Your sourcing team just identified 20 target companies. Someone has to research all of them. That someone used to be a junior analyst with a week and a spreadsheet.
Prompt: "For each of these 20 companies, pull our full interaction history, when we first made contact, who on our team has the strongest relationship, and whether they've appeared in any of our pipelines before."
For each company, the MCP returns every touchpoint your firm has ever had, including first contact date, last contact, meeting history, which lists they've appeared on, and who on your team has the warmest existing relationship. Twenty relationship profiles, loaded in minutes.
Then keep going:
"Which of these have we met with in the last six months but never added to a pipeline?""For the ones where our relationship has gone cold, who should we route the outreach through?"
That last question is where this workflow gets really impressive. Affinity's relationship intelligence scores tell you that a relationship exists, and how strong it is across every person on your team. The answer to "who should make the call" is data-driven, not a guess.
7. The compound query
This is the real unlock. Not a single question, but a working session.
- "Pull our full pipeline with stage, deal size, and deal lead."
- "Flag any deals with no firm contact in the last 30 days."
- "For each flagged deal, who on the team has the strongest existing relationship?"
- "Generate a summary with recommended actions for Monday's partner meeting."
Each prompt builds on what's already loaded. By step four, Claude has the full pipeline in context, knows which deals have gone quiet, and knows who the right person is to re-engage each one. The final output is a prioritized action plan, ready before the meeting starts.
8. And here’s what else is possible…
OK, this is more than one workflow, but what’s above is just a starting point. To wrap up, here are a few more prompts worth trying.
Warm intro finder: "We're about to approach [Company] for a proprietary deal. Who on our team has the strongest relationship with anyone there, and how recently did we last have contact?"
One prompt replaces an hour of "does anyone know someone at..." Slack threads.
LP relationship audit: "Pull all our LPs. Which ones haven't had contact from anyone at the firm in over 60 days? Rank by commitment size."
A prioritized call list, sorted by what actually matters, in seconds.
New partner onboarding: "I just joined the firm. For each deal in our active pipeline, tell me who our primary relationship is with, how long we've known them, who has the strongest connection, and when we last had contact."
The relationship context that used to take a week to absorb, assembled overnight.
Banker relationship scorecard: "Across our pipeline and closed deals, which investment banks appear most frequently? For each, show deal count, average deal size, and when we last had contact with anyone at that firm."
The conversation your partners have been having from memory, now grounded in data.
Conflict check: "We're considering [Company] for a new investment. Do we have any existing portfolio companies or active pipeline deals in the same sector? Has anyone on our team had prior contact with this company or its founders?"
Three lookups and a compliance conversation, collapsed into one.
Frequently asked questions
What is MCP and how does it work with a CRM?
MCP (Model Context Protocol) is an open standard that connects AI assistants to external data sources. In a CRM context, MCP lets tools like Claude, ChatGPT, Gemini, and Copilot read and write your CRM data through natural-language prompts. It can query deal history, pull relationship scores, update pipeline stages, set reminders, and add notes. Instead of exporting CSVs or building custom API integrations, you type a question and get an answer from live data. Affinity's hosted MCP Server requires no engineering setup. Connect it once and your CRM data is available to any MCP-compatible AI assistant.
How can you automate CRM deal management with AI?
Connect your CRM to an AI assistant using MCP (Model Context Protocol), and deal management tasks that used to require dashboards, exports, or analyst time become conversational. Ask "Show me all deals without activity in 30 days, grouped by stage" and get a pipeline health check in seconds. Follow up with "Flag any where the deal lead hasn't emailed the primary contact" and the AI builds on the prior answer. Beyond reading data, AI can write back to your CRM: "Create a note on our open [company] opportunity with today's IC feedback," "Add [company] to our Q2 Pipeline list," or "Pull the call transcript from my last meeting with [company]." This allows you to interrogate your live pipeline data using natural language, getting immediate, nuanced answers about your deal flow without having to build a custom report or wait for a manual audit.
How can you automate CRM relationship tracking with AI?
Relationship decay detection (workflow five) monitors touchpoint frequency across your portfolio and flags relationships going cold, ranked by ownership stake. Meeting briefs (workflow two) pull relationship strength scores and interaction timelines into a single pre-meeting summary. Both turn Affinity’s relationship intelligence data into something you can query conversationally.
What this means for your firm
Eight workflows, all of them built on data that’s already in your CRM. The difference is what sits on top of it: an AI layer that lets anyone on your team ask questions in plain English and get answers in seconds.
Affinity’s latest launch makes this available out of the box. If your CRM data is already in Affinity, these are prompts you can try today.
.png){kind=logo}
