Would You Invest in a Business That Ignores AI?

92 per cent of directors used AI for board work in the past six months. 60 per cent of them sit on boards with no formal AI policy, according to the 2026 OnBoard Board Effectiveness Survey.

Take an investment firm, for example: a new deal lands in the investment lead's inbox, and minutes later a report arrives on why to be excited about it, with a first draft of the investment paper attached. All that is left to do is edit it. 

In this episode of the Tech Transformed Podcast, Shubhangi Dua, Podcast Host and Producer at EM360Tech, speaks with Jamie Roberts, Managing Partner at YFM Partners. YFM Partners is an independent UK-based private equity and venture capital investment firm founded in 1982. 

In particular, they address the key questions: how AI is impacting economic growth, what end-users need and what actions leaders should take to get ahead in the AI race.

They also cover what happens when AI changes an investment decision, why the editing step is where the risk sits, and what a UK board should have in place before AI gets near a vote. 

“AI’s great for a first draft, for the first headline thinking, but then you need the human,” said Roberts.

UK Enterprises Need to Have a Formal AI Policy

86 per cent of limited partners have used generative AI or similar tools to monitor their investments or performance, according to SS&C Intralinks 2026 LP Survey.

The fact remains that a huge number of businesses in the UK still don’t have a formal AI policy, and yet they use AI tools. Dua asks Roberts why that is the case and how, as an end-user of AI tools, the private-equity firm has evolved with the AI developments.  

As a 44-year-old firm, YFM has made over 4,000 investments. 

A large amount of time was initially spent on tidying up historic data before any AI layers were built. “As an old business, we have a lot of history, and some of it was still in boxes, for instance, printed paper in boxes from twenty years ago,” Roberts shared.

“That history gets lost as people leave and retire. For a long time, we relied on the corporate memory of individuals. We started with data in an accessible place.”

They then built an internal AI model around it. This meant they could use large language models (LLMs) without their private data being pulled into them. 

Private Equity AI: Outsourcing Vs Internal Development

The priority when it comes to using an internal development method to keep up with the AI pace is a solid data foundation.

The goal was to make decisions based on 44 years' worth of history and pattern recognition. Without AI capturing 1000s of investments worth millions of pounds would have been incredibly time-consuming and infeasible. 

After building their internal AI model, the YFM managing partner said new opportunities are now run through their AI model automatically. It generates reports listing the reasons to either invest or not invest in the business in simple terms. 

YFM sees about 1,500 opportunities a year and historically converts 1 per cent-  that's around 15 investments. “Every opportunity that comes in gets run through (the internal AI model) first,” Roberts says. One deal would have been discarded had the AI model not reminded the team it knew of probably the world’s leading expert in that space, someone Roberts had met a couple of years earlier.

When asked why YFM chose not to outsource AI development to third parties, Roberts says that the firm wanted the sensitive financial information to remain in-house. While they could have brought in consultants or “bought off-tlf software,” Roberts believed it best to keep the talent in-house. In the beginning of 2026, they hired an AI-native developer to build the internal AI model. 

Buying has a data cost, with a basic off-tlf subscription, “your data risks being in somewhere that you don’t know where it is,” Roberts said.  He advises enterprise leaders to sort their data governance for AI as a priority. 

How to Tackle AI Slop Across Enterprises

A downside of AI, however, Robert says, is that AI churns out too much low-quality content, also called “AI-Slop.” 

YFM manages about £1 billion of other people’s money, so every decision needs an audit trail, and each investment paper follows four structured stages. 

The private equity firm’s leader explains that "AI slop" is a major challenge for the company because their investment decisions require rigorous audit trails and deep analytical thinking. 

When investment leads rely too heavily on AI-generated drafts and simply edit them rather than starting from scratch, they risk bypassing critical deep thinking and end up just "marking the homework."

“How do we make sure we don’t lose the left-hand brain thinking, when all they’re really doing is marking the homework out of our AI model?”

Ten months into using Claude, Roberts feels it is reaffirming what he already thinks. “We need to validate everything that comes out of AI today.” Checking AI’s answers has become a digital literacy skill of its own.

Dua asks Roberts what an enterprise needs before AI can help it grow safely. He says to start with data protection. “If you put data into AI without locking it down, you have no idea where it’s going.” 

His first question for any leader is what data the tool is given and where it ends up, the ground that AI search governance policies cover for enterprises. He does not think a business needs an in-house AI expert. Find the people who use AI for fun, he says, because they build things without being asked.

Enterprise AI as a “Second Brain” 

Dua cites British Chambers of Commerce figures: 54 per cent of UK firms use AI, but only 19 per cent of small business owners feel very confident with it. “Does readiness change what YFM will pay?” 

Roberts says it works the other way round. “The management team that says it’s definitely not going to impact my business, they’re quite hard to back.”

YFM’s investment committee has five members and decides by majority vote. “I can’t see a time anytime soon that we’ll let AI make decisions,” Roberts says. He wants AI in the discussion, as a “second brain” built from the firm’s most experienced people rather than any one partner, and the first step is recording committee meetings. He would be surprised to have it by this time next year. The vote stays with five people, and human-in-the-loop oversight stays with it.

Overall, YFM uses AI to hold 44 years of history in one place and to look harder at deals it would once have dropped. It checks everything AI produces, and it worries more about the person editing a draft than about the draft.

Key takeaways

• YFM has made about 4,000 investments in 44 years.
• Their internal AI model runs every new opportunity before a person sees it.
• About 15 of YFM’s 1,500 yearly opportunities become investments.
• Their internal AI model flagged a world-leading expert YFM had forgotten it knew.
• Editing an AI draft takes less thinking than writing one.
• Roberts uses AI for first drafts and validates everything.
• A management team that dismisses AI is hard to back.
• Roberts’ advice to CEOs: start with one small problem, like buying a car.

For more information, visit https://yfmep.com/.