Skip to main content

Immediate Value

By the end of this tutorial you will have run a whole private-capital workflow — a company signals scan, a reference check, a portfolio or LP pipeline review — from one plain-language request in the AI tool you already use, and set one of them up to fit the way your firm works.

Prerequisites

  • Affinity MCP connected to your AI tool (Claude, ChatGPT or another client). If it isn’t, start with Connect AI Tools to Affinity with MCP.
  • Scale, Advanced, or Enterprise.
  • Web search turned on in your AI tool for the skills that check public news (Company Signals, Portfolio Check-in, Founder References, Market Map). Without it, those checks show as “couldn’t check”.
  • The lists your workflows already use — a deal pipeline, a portfolio list, an LP fundraising list. The skills read them; they don’t need new ones.

Quick-Start Roadmap


Why Skills Matter

More of your team’s work happens in an AI tool than in any one app. A skill packages a whole Affinity workflow — the instructions, the context, the steps — so your AI tool can run it from a single request, using your firm’s own relationships and deal data instead of the open web alone. You don’t need to name a skill or write a prompt. Ask for the outcome, and your AI tool picks the right one. Every skill runs with your permissions and sees only what your Affinity account can see. Most are read-only. The ones that can write — a note, a reminder, a field — show you what they’ll do and wait for your confirmation.

Task 1: Find the Skills and Copy a Ready-Made Prompt

Context

Affinity keeps a library of its skills inside Ascend, with an example prompt for each, so you never start from a blank chat.

Action

  1. Open Ascend → Agents and scroll below the agents to Affinity MCP Skills.
  2. Use the Claude or ChatGPT button next to the heading to open Affinity’s connector in that tool.
  3. Featured skills come first. Click View all to see the library grouped by job: Deals and diligence, Portfolio and LPs, and Data and operations.
  4. Pick a skill, click the copy button next to its example prompt, paste it into your AI tool, and replace the text in square brackets.

Expected Outcome

You know where every skill lives and have a prompt ready to paste.

📚 Help Center


Task 2: Sourcing — Scan Companies for Signals, Map a Market

Context

Two skills help you decide where to look. Company Signals finds recent, dated public news for a set of companies. Market Map lays out a thesis or space and shows how well each company fits.

Action

Company Signals
  1. Ask: “What’s new at Northwind and Brightline this month?” You can give company names, domains or an Affinity list link. The window defaults to 30 days.
  2. Read the signals — launches, funding, executive changes and acquisitions, plus optional partnerships, IPOs, press, earnings and competitors — each with a source link.
  3. Open the sources for anything you’ll act on. It doesn’t use LinkedIn, X or paywalled sources, and it’s read-only.
Market Map
  1. Ask: “Build a market map for vertical AI in logistics.”
  2. In Claude, it renders an interactive map that scores how well matching companies fit your thesis, using Affinity’s semantic search. You can add companies by hand.
  3. If it found companies on the web that aren’t in Affinity yet, it can add them — after you confirm.

Expected Outcome

You have a dated, sourced view of what’s changed at the companies you care about, and a map of the space you’re working.

📚 Help Center


Task 3: Diligence — Find References on a Founder

Context

Founder References finds who can vouch for a founder, executive or candidate, and what your firm already knows about them — your team, early investors, customers, former colleagues and what’s already on file.

Action

  1. Ask: “Who can give me a reference on the CEO of Northwind?”
  2. Review the people it suggests and the evidence for each. It can produce an interactive HTML report.
  3. When you want to reach out, accept its hand-off to Ranked Intro Paths to find the best route and draft the ask.
Founder References is read-only and never sends anything. Ranked Intro Paths drafts outreach on request, and never sends without your explicit approval of the message.

Expected Outcome

You have a short list of references and the warmest path to each.

📚 Help Center


Task 4: Portfolio and LPs — Who’s Due to Raise, Who Needs Attention, What Changed

Context

Three skills cover the recurring questions after the investment: which companies are likely to raise again, which ones your team has gone quiet on while news is happening, and how your own fundraise moved this week.

Action

Ready-to-Raise Indicator
  1. Ask: “Which of my portfolio companies are due to raise?” Give it an Affinity list, or paste domains; each company needs a domain.
  2. Choose the model — Runway (default 8 months), Calendar, Survival or Custom — and the output: a spreadsheet, an HTML snapshot, calendar reminders, or a write-back to existing fields, which happens only after you confirm.
Portfolio Check-in
  1. Ask: “Which portfolio companies need attention?” It asks for your portfolio list, its owner and status fields, and a lookback window (default 30 days).
  2. Read the groups: Needs Attention, Possibly Needs Attention and Couldn’t Check, then Handled, Quiet and No Logged Relationship. Each company takes about 30–60 seconds, so for more than 30 it suggests batches.
LP Pipeline Brief
  1. Ask: “What changed in our LP pipeline this week?” Pick 7, 14 or 30 days, or a date.
  2. Read the new LPs, commitments, stage changes, losses and who needs attention. It works best on an opportunity list with one row per LP per fund.
  3. It can add a note or reminder — only after you confirm.
Ready-to-Raise never predicts a cash-out date. Treat every flag as a prompt to check in, not a forecast.

Expected Outcome

You’ve run a portfolio or LP review in minutes, from data your team already keeps in Affinity.

📚 Help Center


Task 5: Keep Your Data Honest — Audit the CRM, Review Deals You Passed On

Context

Two skills look backwards. Data Quality Audit finds the clutter that makes filters and reports unreliable. Passed Deal Review shows which companies you passed on or lost that later raised.

Action

Data Quality Audit
  1. Ask: “Audit our CRM.” or “Just audit our list fields.”
  2. Open the Excel workbook and the one-page HTML summary of top fixes. It looks at your lists, fields, entries, companies and people — including duplicate fields, duplicate dropdown options and near-duplicate list names.
  3. Make the fixes yourself; the skill is read-only. For duplicate company or person records, use Settings → Duplicate Management (see Tutorial 11).
Passed Deal Review
  1. Ask: “Who did we pass on that just raised?” and choose the period.
  2. Read the anti-portfolio report, broken down by pass reason, industry, geography and team. It needs a pipeline list whose Status field has passed or lost statuses.

Expected Outcome

You have a short, prioritized list of data fixes, and an honest look at the deals you let go.

📚 Help Center


Task 6: Make a Skill Your Own

Context

Every firm runs these workflows a little differently. You can fit a skill to yours in two ways, and the difference matters: one keeps getting Affinity’s improvements, the other doesn’t.

Action

Way 1 (recommended): build a scheduled task on the skill. Ask your AI tool — for example Claude — to run the skill on a schedule, check extra sources such as Slack or Notion, and format the result the way your team wants. For example: “Every Monday at 8am, run the LP Pipeline Brief for the last 7 days and post it in the format we use for partner meetings.” The task loads Affinity’s skill each time it runs, so it keeps getting our updates. Way 2: copy the skill into your own. Ask your AI tool to copy the skill into a skill of your own, then read and edit every instruction. A copy no longer gets Affinity’s updates — you own it from then on.

Expected Outcome

One workflow now runs on your schedule, in your format, without anyone remembering to ask.

📚 Help Center


Good to Know

  • Permission-scoped. A skill only sees and changes what your own Affinity account can.
  • Confirm before writing. Skills that can write show what they’ll change and wait for you. Deleting records, notes and fields, and every merge, always asks first.
  • Skills and agents are different. The Ascend agents (Meeting Prep, Data Update, Warm Intro, Document Synthesis) run inside Affinity. These skills run in your AI tool, on demand, over MCP.
  • Availability: Scale, Advanced and Enterprise, with Affinity MCP connected.

What’s Next

Put a skill on a schedule for the review you run every week, then share the prompt with your team. For the full list of skills — including Meeting Prep, Ranked Intro Paths, Data Migration and Event Setup — see Affinity MCP skills. Affinity AI can make mistakes. Review what a skill produces before you rely on it.