Claude for Private Equity: Automating Portfolio Reporting and IC Memo Drafts With MCP

A common pattern in mid-market private equity: each quarter, an associate spends several days collecting management accounts from every portfolio company, each in its own format, rekeying them into the fund's template, and drafting the performance section of the IC update. By the time the draft reaches the partners, the work that needed their attention, what to do about the companies behind plan, has had the least time spent on it.

Claude, connected to your data through MCP, can take on most of the collection and first-draft work. This guide covers how that workflow is put together, what has to be in place first, and the controls an investment committee will reasonably ask about. For the wider picture across sourcing, diligence and integration, see our guide to how PE and VC firms use Claude.

What MCP Changes for PE Reporting

MCP, the Model Context Protocol, is an open standard from Anthropic for connecting AI assistants to other systems. In practice it means Claude can be given access to the places your portfolio data already lives, such as a SharePoint or Google Drive folder per portfolio company, a data room, a data warehouse, or a financial data provider, and read from them directly.

That matters because the slow part of portfolio reporting is not writing. It is fetching and normalising. Before connectors, an associate had to download files and paste figures into a chat, which was slow and created copies of confidential data in places nobody tracked. With connectors, the assistant reads the source in place, within permissions you control, and every figure it uses can point back to a file and a cell.

Anthropic's financial services offering also includes connectors to data providers and warehouses, which is useful for public comparables and market context. The core of portfolio reporting, though, is your own companies' management accounts, and those usually sit in shared folders.

The Workflow: Portco Data to First Draft

1. Collection. Portfolio companies upload monthly or quarterly packs to a set folder by an agreed date. Nothing new for them except the deadline.
2. Extraction. Claude reads each pack and maps the figures to your standard KPI template, using a mapping document per company that says which line in their pack feeds which KPI.
3. Validation. It checks the extracted figures: totals that should agree, movements outside a set range, missing KPIs. Anything that fails goes on an exceptions list for the associate.
4. Drafting. Using a Claude Project that holds the memo template, the KPI definitions and last quarter's memo, it drafts the performance summary for each company.
5. Review. The deal team reviews the draft and the exceptions, adds the judgement, and owns the final memo.

Steps two to four are where the time goes today, and where AI helps most. Step five should take longer than it does now, because that is where the value is.

Standardise the Template Before You Automate

Automation copies whatever inconsistency it is given. If one company reports EBITDA before management charges and another after, the AI will compare them side by side without complaint. So the first piece of work is not technical at all. Agree a standard set of KPIs and a short definitions document covering revenue recognition basis, EBITDA adjustments, net debt, cash conversion and headcount.

Then build a mapping for each portfolio company, once, showing how their reporting lines translate to your definitions. This is tedious, and it pays back every quarter afterwards. It is also the document your auditors and LPs will want to see if they ask how a reported figure was derived.

Drafting the Performance Summary

With the template, definitions and mappings stored in a Claude Project, the drafting instruction for each company can be short and consistent:

Using the mapping for [Company] and its Q3 pack in the connected folder, complete the standard KPI table for Q3 against plan and Q2.

Then draft the performance summary in the house format from last quarter's memo: revenue and EBITDA against plan, the three largest movements, cash and covenant headroom, and items management has flagged in their commentary.

Use only figures from the pack and drivers stated in management's commentary. Where a movement over 5% has no explanation in the commentary, write [ASK MANAGEMENT]. Cite the file and tab for every figure.

The [ASK MANAGEMENT] tag turns a gap into a question for the next portfolio call rather than a guess in the memo. The citation requirement means a partner who doubts a number can check it in seconds. The same approach works for variance commentary at any level; our guide to checking AI-written commentary sets out the review method in detail.

The Controls an IC Will Expect

• Business plan, not consumer. Claude Team or Enterprise, with training on your data off and retention set by policy.
• Read-only, least access. Connectors see the reporting folders, not the whole drive, and cannot write back to source files.
• Traceability. Every figure in the draft cites its source file and tab, and the mapping document is versioned.
• Exceptions, not silence. Validation failures and unexplained movements are listed, never smoothed over.
• Human ownership. The deal team signs the memo. AI prepares; people decide.
• Confidentiality scope. LP information and live deal material stay out until compliance has approved the set-up.

Where to Start

Pick one fund and one quarter that has already been reported. Build the KPI template and mappings for its portfolio companies, connect the reporting folders, and have Claude produce the performance section. Compare it with the memo the team actually wrote. The differences show you where the mappings need work and where the drafting instructions need tightening.

Run the next live quarter in parallel with the manual process, then switch. Expect the first quarter to take as long as before, because you are building the mappings. The saving arrives from the second quarter on. For value-creation uses beyond reporting, see our guide to AI for private equity value creation.

Frequently Asked Questions

We set up portfolio reporting workflows like this for PE firms and their portfolio companies. Our AI audit and assessment scopes it against your current process, our AI consulting team builds the mappings, connectors and controls, and our AI for Finance Leaders course trains portfolio company finance teams to work with it.