Private equity isn’t just experimenting with AI. From sourcing targets to assessing risks, the technology is moving into the core workflow.
getty
At one of the world’s biggest private equity firms, the pace of work has shifted from weeks to hours. Analysts who once spent long nights parsing financials and industry filings now run the same diligence in a fraction of the time with the help of generative AI. Inside Carlyle Group, the tools have become part of daily life, touching everything from research to credit assessments.
“Ninety percent of our employees use tools like ChatGPT, Perplexity and Copilot,” said Lucia Soares, Carlyle’s chief innovation officer, in an interview with Business Insider. The change, she explained, means credit investors can “assess a company in hours” instead of weeks. In an industry where speed and accuracy define competitive advantage, that’s more than an efficiency play — it’s a fundamental rewiring of how private equity (PE) operates.
And Carlyle isn’t alone. A Bain & Company survey of firms managing $3.2 trillion found that while only a minority have scaled generative AI across portfolios, nearly 20% already report measurable value from deployments. Looking ahead, 93% expect material gains within three to five years. In other words, AI is moving from pilot projects to strategy and becoming the newest entry in private equity’s playbook.
Data-Driven Deal Origination
Deal origination has always been a search for signal amid noise. AI is making that search sharper. For Gelila Zenebe Bekele, founder of Aone Partners, the change has been transformational.
“When sourcing proprietary deals that are not on the market, you either rely on deep industry connections, what we call ‘river guides’ in the search fund ecosystem, or you capture digital signals to identify readiness to transact,” Bekele told me. “Two years ago, some M&A workflows would take a week to complete. Today, an in-house AI system can do it in an afternoon.”
Her vantage point is distinctive. The search fund model — conceived at Stanford GSB in 1984 — was built lean, enabling faster adoption of new tools. That structure has since grown into a robust ecosystem and, by Stanford’s analysis, generated more than $10 billion in value for investors. For Bekele, the lean model creates space to embed AI directly into workflows instead of layering it onto legacy systems.
The market is also industrializing these capabilities. Startups are racing to build connective tissue between unstructured information and investment decisions. One example is Metal, which recently raised $5 million to create an “operating system” for private markets, promising to boost inbound deal flow by as much as 300% without additional headcount.
Rethinking Due Diligence
If sourcing is about finding the right door, diligence is deciding whether to walk through it. Here, Bekele has built AI directly into the process.
“Has there ever been a more opportune moment to modernize operations with technology? In just the past two years, models like GPT, Gemini, and Claude have evolved from simple chat interfaces into agentic systems capable of executing multi-step processes with minimal human oversight.” She added: “And this is only the early innings. By 2030, OpenAI projects AI agents could be tackling problems as complex as drug discovery. Across a wide set of work tasks, generative AI can cut average completion times by more than 60 percent. For technical work, the savings can reach 70 percent.”
At Aone Partners, she has trained AI agents on her standard workflows to generate AI exposure and diligence reports. “The first step is assessing whether the company’s core intellectual property is truly defensible — whether it constitutes a moat that an AI-native startup could not easily replicate.” The second is identifying how data and AI become levers for value creation, from product differentiation to workforce augmentation.
This lens resonates across the industry. Carlyle’s Soares described a structured rollout that trains employees from day one, creates an AI champions’ council and layers proprietary datasets into generative AI — “saving investors from sifting through endless materials.”
The Implementation Questions
Momentum doesn’t erase the risks. For private equity firms, ensuring data security is paramount. “Any AI strategy must be built to ensure that this information remains protected,” Bekele said. Beyond security, she sees a practical challenge many managers share. “It seems to me that every fund manager I speak with is asking the same question — do you build, buy, or partner?”
The pace of technological change complicates that decision. “The reason so many are ‘piloting’ AI for M&A tools is because what feels cutting-edge today may be obsolete in twelve months. The systems I used a year ago don’t compare to what I run today. To me, the answer is building a workflow around the needs of the fund manager that can move in lockstep with the best large language models. If it can be built in-house, fantastic. If it’s a product, the question is whether it can evolve as quickly as the large LLMs themselves.”
Another constraint borders on data quality. PE analysis is nuanced, and generic models won’t suffice. Bain’s research notes that firms making progress are the ones putting in organizational support — standing up governance, prioritizing use cases and spreading learnings across portfolios — rather than experimenting in silos.
What Comes Next
The mood in the industry has changed from curiosity to strategy. Nearly two-thirds of private equity firms now consider AI implementation a top strategic priority, according to Private Equity International’s Advanced Technologies & AI Report.
For Bekele, the reason is simple. “The M&A workflow generates massive volumes of data each day—financial statements, contracts, CRM records, customer reviews and interview transcripts. AI expands the aperture, processing information at speed so investors can focus on what matters: generating insight and making decisions.”
That captures the wager many investors are making: Not that AI will replace human judgment, but that it will elevate it. And it could be the silent revolution happening in an industry known for its strict rules. The firms turning AI into processes, not just projects, will likely compound advantages across sourcing, diligence and portfolio operations. Given how quickly the toolset is evolving, the edge may belong to managers — large or lean — who are open to change and learn in public.