PRIVATE EQUITY WORKFLOW GUIDE

How to Do CIM Analysis With Claude: Screening Deals and Drafting the IC Memo

The CIM tells a clean growth story. Your job is finding exactly where it stops holding up.

This guide walks through how private equity associates use Anthropic's Claude to run a first-pass CIM screen against a fund's investment thesis, flag customer concentration and quality-of-revenue issues with page citations, build the banker question list, and draft IC memo sections that mark what still needs verification. It also covers where an AI read of a CIM breaks down, and why the underlying investment judgment stays with the analyst.

CIM analysis with Claude means using Anthropic's Claude to read a confidential information memorandum alongside a private equity associate, not in place of one. Claude can summarize the business overview, extract customer concentration and revenue mix figures into a table, list every claim that lacks supporting data with a page reference, and draft a first pass of IC memo sections for a human to verify against the data room. It speeds up the mechanical first read; the investment judgment on whether a deal clears the fund's thesis still belongs to the deal team.

Published

Key Takeaways

  • Claude can read a full CIM in minutes and produce a structured summary, but the fund's investment thesis and risk tolerance still have to come from the associate, not the model.
  • The highest-value first pass is a concentration and quality-of-revenue check: pulling every customer, contract, and churn figure into a table with a page reference for each one.
  • A CIM-to-IC-memo workflow works best when Claude drafts against a template the fund already uses and marks unverified claims, rather than writing a memo from scratch.
  • The banker question list Claude drafts from CIM gaps is only useful if someone checks it against diligence already completed, so it doesn't repeat solved questions.
  • Treat every number Claude pulls from a CIM as a citation to verify against the data room, not a fact to carry into the model untouched.

Where Claude Helps in CIM Analysis vs Where the Associate Stays in Charge

A rough split of the CIM-to-IC-memo workflow between what a model can draft and what a person has to own.
TaskClaude's RoleAssociate's Role
First-pass CIM screenReads the full memo and summarizes business model, market, and financials against the thesisDecides whether the thesis fit is real or a surface-level match, and whether to proceed
Customer concentration and quality-of-revenue flagsExtracts every relevant figure into a table with page references and notes unsupported claimsJudges how much concentration risk the fund can tolerate for this deal size and sector
Banker question listDrafts questions tied to specific CIM gaps and page numbersPrioritizes which questions actually matter and removes ones diligence has already answered
IC memo draftingWrites a first pass of memo sections in the fund's format, marking unverified claimsRewrites the investment recommendation and owns everything that goes to the committee
Final investment decisionHas no roleMakes the call, based on judgment no model can substitute for

Copy and Paste

Prompts You Can Use Today

Screen a CIM against your fund's thesis

You're a private equity associate helping me do a first-pass screen of this CIM against our fund's investment thesis, which targets founder-led B2B software companies with recurring revenue above eighty percent of total sales, healthy EBITDA margins, and a clear path to a platform build-out. Read the attached CIM and tell me, section by section, where the business matches the thesis, where it falls short, and where the memo simply doesn't give you enough information to tell either way. Cite the page number for every claim you reference.

Swap in your fund's actual thresholds and sector focus before running this.

Flag customer concentration and quality-of-revenue risk

Go through this CIM and pull out every data point related to customer concentration, contract length, revenue recognition, and churn. Build a table with columns for the metric, the number reported, the page it appears on, and a flag for whether the memo explains the methodology behind that number or just states it. Call out any place where a growth or retention figure is asserted without showing the underlying calculation.

Ask for the table in a format you can paste straight into your working notes.

Build the banker question list

Based on the gaps and ambiguities you found in this CIM, draft a list of specific questions for the sell-side banker, organized under headings for financial diligence, commercial diligence, and legal or regulatory diligence. Each question should reference the exact page or exhibit in the CIM that prompted it, and should be phrased the way an associate would actually ask it on a call, not as a generic diligence checklist item.

Cross-check the list against questions already answered before sending it.

Draft the first section of the IC memo

Using this CIM and the concentration and quality-of-revenue notes we already pulled together, draft the business overview and investment thesis sections of an IC memo in the voice and structure our fund typically uses: a two-paragraph summary of the business, three to five bullet points on why this fits our strategy, and a short list of open diligence items. Mark every sentence that relies on a CIM claim we haven't independently verified with a bracketed note.

Feed it your fund's actual memo template so the structure matches on the first draft.

Stress-test the growth story

Read the growth projections and market sizing sections of this CIM and walk me through the assumptions the sponsor is making to get from current revenue to the projected exit multiple. Identify which assumptions are supported by data elsewhere in the memo, which are asserted without support, and which depend on market conditions the CIM doesn't actually demonstrate the company controls.

Use this before the banker call, so unsupported assumptions turn into questions.

From Anthropic

What Anthropic offers for Private Equity

Advancing Claude for Financial Services

An October 2025 update adding Agent Skills for comparable company analysis, discounted cash flow models, due diligence data packs built from data room documents, and company teasers, plus new connectors including Chronograph for portfolio monitoring and Egnyte for data rooms.

· Source

Agents for financial services

Ten agent templates released in May 2026, including a pitch builder, model builder, valuation reviewer, and market researcher, each available as a plugin in Claude Cowork and Claude Code. Anthropic also made its Excel, PowerPoint, and Word add-ins generally available.

· Source

In Practice

How Private Equity Teams Use Claude

According to Anthropic's published customer story

Hg

Anthropic's May 2026 financial services post quotes the head of Hg Catalyst describing Claude as a powerful tool for automating complex analysis.

· Read Anthropic's customer story

According to Anthropic's published customer story

Chronograph

According to Anthropic's customer story, Chronograph, which builds private equity portfolio monitoring software, reached full adoption of Claude across its staff with high daily usage, and new hires became productive within weeks instead of months.

Read Anthropic's customer story

Illustrative walkthrough

The First-Pass Screen

  1. An associate pastes the fund's thesis and the CIM into Claude and asks for a section-by-section read against that thesis.
  2. Claude returns a summary flagging where the business matches the thesis, where it falls short, and where the memo simply doesn't say enough to tell.
  3. The associate uses that summary to decide, within the same afternoon, whether the deal is worth a first call with the banker.
  4. The thesis fit judgment and the go or no-go decision stay with the associate; Claude only organized the read.

Illustrative walkthrough

The IC Memo Draft That Wasn't Ready

  1. A deal team asks Claude to draft the business overview and risk sections of an IC memo from the CIM and the diligence notes gathered so far.
  2. The draft reads cleanly, but several sentences carry a bracketed flag noting a claim that traces back to the CIM itself rather than independent diligence.
  3. The deal lead walks through each flagged sentence against the data room, cutting claims that data room documents didn't support and rewriting the risk section in their own words.
  4. The memo that reaches the committee reflects the deal lead's judgment, with Claude's draft as scaffolding rather than the final text.

Private Equity and Claude: the Published Numbers

Figures as published by Anthropic, linked to each source.
FigureContextSource
100% adoption across 150+ employeesChronograph, a private equity software company, per Anthropic's customer storySource
80% daily active usageChronograph's reported daily usage of Claude, per Anthropic's customer storySource
new hires productive by week 2 vs 2+ months previouslyChronograph's reported onboarding change, per Anthropic's customer storySource

THE WORKFLOW

What Claude Actually Does With a CIM

Four tasks account for most of the value in a CIM-to-IC-memo workflow.

First-Pass Summary

Claude reads the full CIM and produces a structured summary of the business, market, and financials, so the associate starts diligence with a map instead of a blank page.

Concentration and Quality-of-Revenue Flags

It pulls every customer, contract, and churn figure into a table with a page reference, and flags claims the CIM makes without showing the underlying math.

Banker Question Lists

Gaps and ambiguities in the memo become a draft list of diligence questions organized by financial, commercial, and legal topic, each tied to the page that prompted it.

IC Memo Drafting

Claude can draft the business overview, thesis fit, and open-items sections of an IC memo in the fund's own format, marking any sentence that still needs verification.

STEP BY STEP

How to Analyze a CIM With AI, Step by Step

The workflow starts before you upload anything. Write down the fund's thesis in plain language: the sector, the check size, the margin and growth thresholds, and the deal-breakers. Claude needs that thesis in the prompt every time, because without it the model will summarize the CIM generically instead of testing it against what your fund actually looks for.

Once the thesis is in the prompt, have Claude work through the CIM section by section rather than all at once. Ask for the business overview and management team first, then market sizing, then the financial package, then legal and regulatory disclosures. Section-by-section prompts produce more specific answers than one long request, and they make it easier to spot where the memo is thin.

The second pass is the concentration and quality-of-revenue check. Ask Claude to build a table of every customer, contract length, churn, and revenue-recognition figure in the memo, with a page reference for each one and a flag for any number that's stated without showing how it was calculated. This is the single highest-leverage use of CIM analysis with AI, because it turns a slow manual page-by-page hunt into a review of a table.

From there, the banker question list and the IC memo draft follow naturally. Ask for diligence questions tied to specific gaps the CIM left open, then ask for a first draft of the memo sections your fund's template covers, with unverified claims marked. Each output should be reviewed against the data room before it goes anywhere near a partner or an investment committee.

SPEED, NOT REPLACEMENT

CIM Screening and Deal Screening AI: Where the Speed Comes From

The time savings in CIM screening come from a few specific bottlenecks, not from the model doing the whole job.

Faster First Reads

A CIM that takes an associate an afternoon to read closely can be summarized and cross-checked against a thesis in a fraction of the time, freeing up hours for the parts of diligence that actually need judgment.

Consistent Structure Across Deals

Deal screening AI applied the same way on every CIM produces summaries and flags in a consistent format, which makes it easier to compare deals side by side instead of relying on however each associate happened to take notes.

Fewer Missed Page References

Because Claude can hold the entire memo in context, it's less likely than a tired reader to miss a footnote on a later page that contradicts the headline growth number up front.

More Time for the Judgment Calls

The hours saved on the mechanical read go toward the calls that actually decide a deal: how much concentration risk is tolerable, whether the market story holds up, and whether the team can execute.

THE HONEST ANSWER

Can AI Replace Private Equity Analysts?

No, and the reason is specific rather than sentimental. CIM analysis with AI is strong at the parts of the job that are mechanical: reading dense text quickly, extracting figures, checking a memo against a written thesis, and drafting a first pass of a document with a known structure. It's weak at the parts of the job that require judgment under uncertainty, because a CIM is written by a banker to sell a deal, and reading past that framing is exactly what an experienced analyst does that a model cannot be trusted to do alone.

The tell is in what a CIM leaves out. A memo can present a clean growth chart while quietly excluding the one customer whose contract is up for renewal, or describe a market as large and underpenetrated without addressing why competitors haven't already taken share. Claude will summarize what's on the page accurately, but it has no independent way to know what a seller chose not to include, and that's often where the real risk in a deal lives.

What changes, realistically, is the shape of the associate's day. Less time goes into the first read and the mechanical extraction of numbers; more time goes into deciding what the CIM is quietly avoiding, pressure-testing the assumptions behind the projections, and owning the recommendation that goes to the investment committee. That shift is a change in where the hours go, not a change in who is accountable for the call.

WHAT'S SHIPPED

What Anthropic Has Built for Financial Services Teams

Anthropic has been building out this category directly rather than leaving it to general-purpose prompting. An October 2025 update to its financial services offering added Agent Skills for comparable company analysis, discounted cash flow models, due diligence data packs built from data room documents, and company teasers, along with connectors to tools like Chronograph for portfolio monitoring and Egnyte for data rooms, according to Anthropic's own announcement.

In May 2026, Anthropic released ten agent templates aimed at financial services work, including a pitch builder, a model builder, a valuation reviewer, and a market researcher, distributed as plugins inside Claude Cowork and Claude Code, alongside general availability for its Excel, PowerPoint, and Word add-ins.

The clearest signal of how this plays out in practice comes from Chronograph, a private equity portfolio monitoring software company: according to Anthropic's customer story, Chronograph reached full adoption of Claude across its staff, and new hires became productive within weeks instead of months. Hg Catalyst's head has separately described Claude, per Anthropic's May 2026 post, as a powerful tool for automating complex analysis. Those are accounts of internal teams, not case studies of a single CIM being analyzed end to end, so treat them as evidence that the underlying capability is real, not as a substitute for testing the workflow on your own deals.

THE OUTPUT

IC Memo AI Drafts: What to Keep, What to Verify

An IC memo AI draft from Claude is a starting point, not a finished document.

Keep the Structure

A draft that follows your fund's usual memo format, section by section, saves real time even if every sentence inside it needs a second look.

Verify Every Marked Claim

Ask Claude to flag any sentence that relies on a CIM assertion your team hasn't independently checked against the data room, and treat that flag as a to-do list, not a footnote.

Rewrite the Recommendation Yourself

The investment recommendation and the risk framing are the parts of the memo that should read in the deal lead's voice and reflect judgment a model can't own.

Keep the Data Room Out of Public Tools

Run this workflow only through channels your fund has approved for confidential deal information, and never paste data room contents into a consumer chat tool.

CONFIDENTIALITY AND SIGN-OFF

The Limits: Confidentiality, Hallucination, and Human Sign-Off

CIM analysis with AI touches some of the most sensitive documents a fund handles. A confidential information memorandum is shared under an NDA, and the target company, its management team, and its customers all have an expectation that the information stays inside a controlled process. Before running any CIM through Claude or any other tool, confirm the workflow runs through a channel your fund's compliance and IT teams have actually approved, not a personal account.

The second limit is straightforward hallucination risk on numbers. Financial figures, page references, and exact wording are exactly the kind of detail a language model can get subtly wrong, especially across a long document with dense tables. Every figure that ends up in a banker question or an IC memo draft needs to be checked against the source page in the actual CIM before anyone relies on it.

The third limit is who signs off. No output from this workflow, whether it's a concentration flag, a banker question, or a memo section, should reach an investment committee, a lender, or a regulator without a human on the deal team reviewing it first. That review isn't a formality: it's the point at which someone with context on the deal catches the thing the model had no way to know.

None of this argues against using AI for CIM analysis. It argues for treating the output the same way you'd treat a first draft from a junior analyst: useful, often accurate, and never final without a second set of eyes that carries the accountability.

Learn to Run Workflows Like This, Not Just Read About Them

Reading about CIM analysis with Claude is different from being able to run a section-by-section screen, build a concentration table, and know which claims to double-check. Nightschool AI's free curriculum is hands-on from the first lesson, so you build that judgment by doing it, not by watching someone else describe it.

Frequently Asked Questions

How do you analyze a CIM with AI?

Give Claude the fund's investment thesis and the CIM, then work through it section by section: business overview, market, financials, and legal disclosures. Ask for a table of customer concentration and quality-of-revenue figures with page references, then a banker question list built from the gaps, and finally a first draft of IC memo sections with unverified claims marked. Verify every figure against the data room before relying on it.

Can AI replace private equity analysts?

No. AI can speed up the mechanical parts of CIM analysis, like summarizing a memo and extracting figures, but it can't reliably tell you what a CIM is quietly leaving out, judge how much risk a deal's concentration profile represents for your fund, or own the recommendation that goes to committee. Those judgment calls stay with the analyst.

What is CIM screening and how does AI speed it up?

CIM screening is the first-pass read of a confidential information memorandum to decide whether a deal is worth pursuing further. AI speeds this up by summarizing the memo against your fund's thesis and pulling key figures into a structured format in minutes rather than an afternoon, so the associate spends more time on the judgment call and less on the mechanical read.

Is it safe to run a CIM through Claude?

Only through a channel your fund's compliance and IT teams have approved for confidential deal information. A CIM is typically shared under an NDA, so it shouldn't go into a personal or unapproved account. Check your fund's data handling policy before uploading any data room material.

What should an IC memo AI draft actually include?

A useful IC memo AI draft follows your fund's existing template and marks any sentence that relies on a CIM claim the team hasn't independently verified. It should not include the final investment recommendation in its finished form, since that section reflects judgment the deal lead has to own directly.

Can Claude catch quality-of-revenue issues in a CIM?

Claude can flag where a CIM states a revenue or retention figure without showing the underlying calculation, and it can pull contract terms and churn data into a table for review. It can't independently verify those figures against source documents outside the CIM, so quality-of-revenue conclusions still need to be checked against the data room.

How is deal screening AI different from reading a CIM manually?

Deal screening AI does the same first read a person would do, but faster and with a consistent structure across every deal, which makes it easier to compare opportunities side by side. It doesn't replace judgment about deal-specific risk; it changes how much of the associate's time goes into the mechanical read versus the analysis.

Does using AI on a CIM change what a private equity associate needs to know?

It shifts the emphasis rather than removing the need for expertise. An associate still needs to know what a CIM typically omits, how to read financial statements critically, and how to judge deal risk. What changes is that less time goes into manual extraction and more time goes into scrutinizing what the model surfaced.

Ready to Run Your Own CIM Screen?

The fastest way to trust the output is to have built it yourself.

Nightschool AI is an independent learning platform and is not affiliated with, endorsed by, or sponsored by Anthropic. Claude is a trademark of Anthropic, PBC. Looking for Anthropic's official Claude Academy? It's at academy.claude.com.