Perspectives · Private Equity
Why not just use ChatGPT or Claude?
A direct answer to the question PE firms ask most when we talk about Compass AI for private equity.
Dave Elfanbaum
Strategic Advisor, Kairoi
April 2026 · 5 min read
The question we hear most
When we talk to PE firms about Compass AI, the most common objection isn’t a competing vendor. It’s this.
“We’ve been using ChatGPT across the team. Is what you’re offering meaningfully different?”
It’s a smart question, and it deserves a straight answer.
A fair question
A generic LLM like ChatGPT or Claude is powerful, and it has no idea your firm exists. It’s never read your deal memos. It doesn’t know your investment thesis, your portfolio companies, your past underwriting calls, or the lessons you’ve learned over twenty years of doing this work. Every conversation starts from zero.
That isn’t a knock on the technology. It’s the right way to set expectations. Generic models are reasoning engines, and they’re powerful when their context is rich and generic when their context is generic. For a firm whose edge lives in judgment built up across hundreds of deals, that limit matters.
What firm-specific intelligence actually looks like
CompassPE is Compass AI set up for private equity. It’s built behind your firewall and connected to your SharePoint. It reads and organizes your real documents, including deal memos, board presentations, financial models, contracts, and executive assessments. It reasons across them.
When a partner asks whether a portfolio company has outstanding debt, the system doesn’t search the internet. It reads the loan agreement, checks the latest balance sheet, looks at the debt schedule, and answers from your own files.
That kind of capability is secure, firm-specific, and grounded in your documents. You can’t get there by asking a generic LLM to help draft something. The grounding isn’t a feature bolted on top of the model. It’s the architecture around it.
Workflows are the second moat
The second difference is workflows.
CompassPE ships with agentic workflows built for private equity. These include CIM analysis, investment thesis scoring, management team assessment, diligence question generation, executive search, and portfolio monitoring. Each one is shaped by what actually happens inside an investment lifecycle, not by what a generalist assistant might guess at.
Building those workflows from scratch on top of a generic LLM takes a serious consulting effort at a large firm. Costs run into hundreds of thousands of dollars. Timelines run into quarters. The work is real, and someone has to do it. Your firm can fund that build from scratch, or you can buy capability that’s already running inside other firms.
What you’ve already built isn’t lost
If your team has already built custom GPTs or written internal prompting frameworks, that work doesn’t have to be thrown out.
We hear versions of this in almost every first conversation. “We’ve already put money into this. We don’t want to start over.” Fair. We can bring what you’ve built into the CompassPE environment, so the earlier work isn’t lost. It gets stronger because it sits inside a system designed to use it.
The frame, plainly stated
CompassPE isn’t ChatGPT with a different logo, and it isn’t a thin layer of prompts wrapped around a generic model. It’s an AI platform built for your firm, designed to know your documents and run the workflows you actually use.
The foundation models, including GPT and Claude and Gemini, are becoming interchangeable fast. The real moat is moving to what wraps around them. Your historical deal notes. Your specific investment thesis. Your portfolio company data. The PE workflows that turn raw model power into something you can act on.
That context takes time to build right. It can’t be bought off the shelf, and the firm that builds it now will compound that lead across every fund cycle that follows.
ChatGPT isn’t bad. What your firm needs from AI is different, and a generic model can’t get you there no matter how nicely you ask.
See it on your own data
The best way to understand what this can do is to see it.
We’ll run a complimentary AI assessment of one company in your portfolio — a structured look at where AI creates risk and where it creates value, delivered in a working session with actionable findings.