Meet Petra, the AI that runs a PE firm

Meet Petra, the AI that runs a PE firm

Madeline Shi

Fri, August 21, 2026 at 10:26 PM GMT+3 6 min read

Jenna O'Malley/PitchBook News

Talk of AI adoption is pervasive in PE, but would you trust the technology enough to let it manage your own firm? Ethos Capital is testing these limits.

The Boston-based PE firm spent around five years building Private Equity Transformation Research Agent—or Petra—an AI assistant trained on more than 50,000 data sources, including decks outlining every deal the firm has taken or passed on, as well as public information ranging from regulatory filings to news and video.

Today, Petra is woven into nearly every aspect of Ethos' business, from filtering pitch decks, running initial diligence, monitoring portfolio companies to managing investor relationships. Ethos even uses AI to track its own employees' work, according to Fadi Chehadé, co-founder and managing partner at the $7 billion investment firm.

Launched by Chehadé and Erik Brooks, two Abry Partners alumni, in 2019, Ethos targets middle-market companies with an enterprise value of $200 million to $2 billion, focusing on supply chain and logistics, insurance and financial services and digital platforms.

The firm manages a $512 million closed-end fund, co-investment capital and a multi-billion-dollar continuation fund recently raised to house Identity Digital, a portfolio company it backed in 2021. The CV has brought in new commitments from big-name managers, including Accel-KKR, TPG, Neuberger Berman, CVC Capital Partners and Coller Capital.

Ethos' AI tool runs the first pass on every pitch that crosses the deal team's desk, scoring a target against a set of preferred metrics, identifying strengths and risks, conducting market analysis and rating the management team. What once took two to four weeks is now compressed to about half an hour, Chehadé said.

Petra also has visibility into the firm's inner workings. It produces a digital footprint of every employee, which helps managing partners like Chehadé to stay on top of what they are doing. It sees every piece of correspondence: every email, every calendar invite and every Slack message and directs the information to the people it determines need it.

"When I log into Petra, it knows me, and it talks to me," Chehadé said. "If I ask it what's happening on a deal, it reads every email, every calendar invite and every note in the entire company and comes back to me with a digest based on who I am."

When asked how his employees take to Petra monitoring their work, Chehadé disputed the phrasing. He said that while the AI system captures work outputs produced by every user, it only shares the data based on "carefully designed rules that reflect privacy issues and access rights."

"Petra is not designed to track individuals' work; rather, work product", such as analysis, notes and models, he wrote in an email. PitchBook did not speak with the firm's employees.

A longtime technology entrepreneur, Chehadé said the tool empowers a deal team composed of only a handful of investment professionals. This small crew reviews 130 to 140 companies a year and moves a dozen to in-depth due diligence.

With the AI system, Ethos has also cut its reliance on outside advisers to some degree. Chehadé said he used Petra to run preliminary research on a target's legal risks in minutes, and the tool addressed an estimated 90% of his questions. The same task would have gone to an associate at an outside counsel and taken weeks. He still uses lawyers today, Chehadé said, but he comes to them with sharper questions.

Sign up for The Daily Pitch newsletter Subscribe

The argument for human touch

The idea of running a PE firm on AI was widely endorsed by others spoken to for this piece. The most avid think the technology's expansion is inevitable, and that service providers across the M&A industry may see AI take over parts of their work and reshape fee structures from hourly-based billing toward outcome-based compensation.

"The cost of due diligence should go way down," said Nishat Mehta, chief executive at Lexitas, a legal tech provider backed by Apax Partners. "If you give me a deal room and throw 10,000 pages of documents there, today I may pay an outside party to review all those documents and surface a set of questions and concerns that I need to pay attention to. AI can probably start to do a decent amount of that work."

Still, few believe AI will fully replace humans in the PE, and no one said they'd trust an AI-made decision outright, though their perceptions of how that collaboration should play out and what they think AI should be trusted to do vary.

Nitin Gupta, the co-chief investment officer at small- and mid-cap PE investment firm Flexstone Partners, takes the PE industry's enthusiasm for AI adoption "with a grain of salt," questioning how far down the chain this implementation actually reaches.

He noticed that GPs are using AI to analyze data, drive efficiencies, inform better decision-making and embed it in portfolio companies for value creation. But he values human judgment built on experience much more highly. AI, he said, can produce "hallucination effects," making decisions based on a few words in a document.

"It's not always just an automatic, quantitative decision," Gupta said. "It's also your gut feel for the company, for the management team, and the value-add your operating partners are bringing to the table. There are a lot of subjective, qualitative factors that go into decision-making."

Flexstone oversees separately managed accounts for institutional investors, investing in middle-market PE funds of up to $2 billion.

Kyle Griswold at FTV Capital, a technology-focused growth equity firm, holds a similar view.

He said AI is good at digesting data and screening for potential targets, but his team still hunts for new business the old fashioned way.

"We can tell when we feel like we're getting AI spam," he said. "We do a ton of hopping on planes and picking up the phone and going to conferences. None of that can be replicated."

AI can produce faster, deeper market research, but the proprietary information that comes through private conversations with competitors and customers cannot be replicated, Griswold said. He wants his deal team to understand accounting and finance, and know how to model a business themselves, even when a tool could do it faster.

Ethos' Chehadé made clear that Petra isn't replacing the human judgment and effort his team brings to a deal.

"I can't trust machines to make these calls because they will make them based on logic but not based on conscience," he said. "We overdo it when we say AI is replacing human bankers, advisers or lawyers; we might be replacing some of their functions."

He said Petra may draft an email to approach a target business, but he requires a human to review and polish that message, and nothing goes out until the person presses send.

Chehadé expects that within the next decade, society will shift from a human-led world to a hybrid one, where machines and humans divide responsibilities.

"In an investment environment, it is important, when we design these things, to make sure humans are ultimately in control," he stressed.

This article originally appeared on PitchBook News