How to build fund infrastructure with AI

What this workshop is about

Fund operations can eat up your time. Portfolio reporting, data collection, limited partner requests and updates, triaging inbound deal flow, compliance and administrative requirements and more take your time away from meeting founders, doing valuable due diligence, and doing the work that drives the returns.

The venture industry has traditionally been underserved by quality software to assist with fund operations: too small, too varied, too specific, outside the scope of the revolution in software to run the operations stack of modern companies. But AI has changed that. Fund managers are actively looking for ways to improve on their processes and build AI-supported processes and tools to improve fund operations.

Most people are still using AI by prompting it one question at a time. This workshop is about the level above that: creating a repeatable way to identify a workflow worth building, design the process and tool so that people trust the output, build the smallest version that works, and keep it running after launch. It's easier than ever to build a new tool, but turning that demo into a process and tool that improves our lives is still hard.

Register for the workshop and you get an invite to the next live session, and also immediate access to the recording of the last live session, slides, and worksheet right away. Join any future live session if you can't make the next one, and schedule a 1:1 with me to go over anything we cover.

What will be covered

The session runs in three parts:

• Product thinking, adapted for AI. Watch your real week before you design anything. Hire AI for a job a person currently does, not "a tool." Define what v0.1, v1.0, and v2.0 mean before you build, so you know when you're done. Ship one workflow, then expand.
• A repeatable framework: identify, design, build, maintain. How to spot the first workflow worth building, against five concrete tests. How to design for trust, with a source reference on every output and a human reviewing the edge cases. The three build shapes (extract-store-query, source-synthesize-review, trigger-classify-act) and how to pick. And the part nobody plans for: keeping it alive so it doesn't quietly rot.
• The technical foundations. The five levels of working with AI tools, from prompts to projects to skills to MCP to agents, and when to move up each rung. Why context, not prompt-craft, is the real skill. Individual vs team accounts. When to appoint a coordinator instead of hiring a builder. How to connect Claude or OpenAi to the data you have in your business.

You'll also get the worksheet I built to frame this for your own fund, and you can schedule a 1:1 with me anytime for practical help on building your tools.

Who this is for

GPs, fund operations people, and the associate who has quietly become the one who builds things. If you run a part of a fund's back office and want to build your own workflows instead of paying platform fees for someone else's, this is for you. It assumes you want to actually build, not just understand the idea in the abstract. No engineering background needed.

What you'll walk away with

A framework you can run on your own first workflow, and the worksheet to scope it before you write a single prompt. You'll know which workflow to start with, how to design it so the output holds up when real money and LP relationships are on the line, which of the five Claude tools fits the job, and what it takes to keep the thing alive after launch.

Related

• Portfolio Reporting, the source-available product I built for fund operations
• Venture Capital Model, the full fund forecast tool
• Fractional CFO, if you want ongoing help beyond the workshop