Welcome to Uncommon Capital. I built RWT Capital from a solo advisory practice in Kelowna into a fast-growing firm with offices in three cities and 130+ mandates under our belt. I share the real lessons from inside the deal to help business owners build value, exit well and get deals across the line.

Not long ago, we worked with the owner of a project-based business tied to new residential construction. It was a niche business with great margins, but highly exposed to the broader economy.

We brought him an offer that topped our estimated valuation range. Then, while we were negotiating the LOI, he backed out.

Why? He had been using ChatGPT for everything M&A related… you name it, including what we should be doing as a firm. Most of what it gave him was flawed because it relied on the wrong data or irrelevant. We could see the real estate market softening and were confident his business was about to see a dip, so we urged him to take one of the deals on the table. It would have accomplished exactly what he came to us for, which was to retire and spend more time with his family. Instead, he built a go-forward strategy he was convinced would double or triple the business in the next few years. He was steadfast that his strategy was rock solid… and we never took an offer to exclusivity.

Our engagement collapsed.

We hate seeing deals not progress. But it was also an early warning sign of what we are now seeing from many of the business owners we work with.

I've been doing this work long enough to know the classic M&A mistakes by heart. Owners who aren't ready emotionally. Financials that tell three different stories. Expectations built on what a friend sold for in 2021 in a completely different industry. Those haven't gone anywhere. But over the last two years, a new category has shown up in our files, and it's growing quickly… mistaken AI tactics being introduced to a deal.

AI has made the easy parts of a deal faster. That's genuinely good and I welcome it. We use it every day at RWT, for research, for analysis and for things I wont share as they are proprietary. But it's also quietly increasing risk in the parts of a deal that were always hard… and most business owners don't realize it's happening.

AI tools make the wrong answer look polished. That's the whole problem.

The first mistake is false confidence on value.

A chatbot will give you a number, usually built from public comparables, generic industry multiples and whatever it can infer from what you've told it. What it can't tell you is how a strategic buyer in your niche will view your customer concentration, whether a private equity group will discount you for key person risk, what your capex profile does to value, or what the debt markets look like the month you go to market.

And most owners have nothing to check it against. The Exit Planning Institute found that only 27% of Baby Boomer business owners have ever completed a formal valuation, even though more than half plan to exit within five years. When the only number an owner has seen comes from a chatbot, that number becomes the truth.

The number itself isn't the danger. The danger is the psychological anchor it creates, one the market will never pay. An owner anchored to a figure treats every real offer as a lowball. Negotiations that should be collaborative become adversarial. Good buyers walk away quietly, because they can sense a seller who will never close.

The second mistake is polished documents that do more harm than good.

Letters of intent, term sheets, even full CIMs and pro forma financials drafted with AI look clean. They read well. They have all the right headings. We now regularly see owners arrive with a complete exit strategy: the timing, the sequence, who will buy them and for how much. Often it was fifteen minutes on ChatGPT with incomplete data underneath it, and it's full of fundamental errors. That matters, because what a seller puts in front of a buyer can come back to haunt them long after closing. Information accuracy is critical, and so is the strategy behind it. When both have major holes, the seller is exposed.

Working capital pegs are a good example. The peg is one of the most consequential numbers in a transaction, and it can move the real purchase price by hundreds of thousands, or even millions, of dollars. A template will fill it in with a reasonable-sounding method. Reasonable for whom?

Earnouts are worse. A clause that seems straightforward on signing becomes a dispute two years later, when the buyer has changed how revenue is recognized and the seller's payout quietly disappears. According to SRS Acquiom, earnouts pay out roughly 21 cents on the dollar on average and are contested at least 28% of the time. An earnout clause generated by a template is a promise that will most likely never be kept.

The third mistake is speed without readiness.

Buyers have adopted AI at scale. A 2025 Deloitte survey of 1,000 corporate and private equity leaders found that 86% have already built generative AI into their M&A workflows, most of them within the past year, and more than a third use it specifically for screening targets and due diligence. Diligence that used to unfold over weeks now arrives in a single wave: hundreds of questions, generated and organized in an afternoon, all expecting answers.

A seller who hasn't prepared for that volume gets overwhelmed. Answers come back late, inconsistent or incomplete. Worse, they often don't match what was portrayed in the AI-driven documents. The buyer starts to wonder what else is missing, and trust erodes at exactly the point where it matters most.

Axial's 2025 Dead Deal Report shows how this plays out. Diligence findings are now the leading reason signed LOIs fall apart, and quality of earnings discrepancies alone doubled as a cause of failure between 2023 and 2025. The deal that dies in diligence often dies because the seller couldn't keep up to the magnitude of demands being placed on them.

The fourth and perhaps biggest mistake is tactical mis-judgement.

The first three mistakes aren't new. They are risks on any transaction, but AI is severely heightening them. The bigger risk is that AI is now making tactical calls for sellers, based on incomplete data and a limited understanding of the human complexity of a mid-market deal.

That's what happened to the owner I opened with. His AI-driven strategy ignored the one risk every business owner faces… exposure to the economy. And he isn't alone. AI tactics tell sellers they have unlimited offers waiting and should act accordingly, when in reality they may only have a few. They are static decisions in a highly dynamic environment, driven heavily by whatever the user puts in.

That last point is backed by hard research. A Stanford study published in Science in March 2026 tested 11 leading AI models and found they affirmed users 49% more often than humans did. People who got that agreeable advice became more convinced they were right, and couldn't tell an AI that was flattering them from one being objective. In other words, AI tends to tell a seller what they want to hear… and the seller walks away more certain than ever.

As a firm, we spend more and more time re-educating clients and prospects on what they have “learned from AI.” We're willing to be honest with them, even at the risk of losing the client entirely. But what happens when an owner has no one to be honest with them? Who tells them that treating a buyer as dispensable will sink the relationship, that an all-cash offer is unrealistic for most mid-market companies, or that they won't be walking away from the business the day the deal closes?

What happens is that the tactics they have deployed start to crumble around them.

What AI can't do

None of this is an argument against the technology. AI belongs in every deal now, including ours. It makes good advisors faster and better informed, and the firms that refuse to use it will fall behind.

But it is a tool. Think of a surveyor's total station… it can do remarkable things for a surveyor, but in the hands of someone who doesn't know how to use it, it means nothing. AI in M&A is the same.

A deal isn't won on the speed of the analysis. It's won on judgment, on timing, and on the human moments where everything nearly falls apart. Knowing when to push and when to wait. Reading the room when a buyer's tone shifts. Recognizing that a seller's objection to a clause is really about something else entirely.

We like to say deals die twice before they close. It's usually late. It's usually a phone call. Someone has lost their nerve, or their trust, or their patience, and the whole thing is hanging by a thread.

AI isn't on that call. It never will be.

For the business owners reading this and using AI, don't mistake a faster process for a better one. The mistakes are new… but the cost of making them is the same as it's always been, perhaps even higher in an economy that's changing this quickly.

As for the owner I started with… his business saw a big pullback as demand dried up. What we saw coming… came.

-

Reece Tomlinson is the Founder and CEO at RWT Capital Corp. and the author of Uncommon Capital.

‍