Family Business Succession in the Age of AI: Keep It, Hand It Down, or Sell?

Family business succession planning is the work of deciding who will own and who will run the business after you, and getting the company ready for that person before the handoff is forced. Done well, it starts years before anyone retires and answers two separate questions: who holds the equity, and who holds the keys. Most of the trouble comes from treating those as one question.

I want to make an argument that the standard advice on this has gone stale in about eighteen months, and that the people most eager to give you that advice right now are buyers.

The moment you are in

Here is the company I am writing to. It does something skilled and physical: industrial services, specialty contracting, a clinical practice, logistics, staffing, an accounting or insurance practice. Somebody's parent or grandparent founded it. It has never taken outside equity, so nobody has ever told it what it is worth except a customer. Its people are excellent, its systems are twenty years old, and the customers have stayed for decades because the company knows things about them that are written down nowhere.

And the founding generation is handing it over, or trying to, and the kids may or may not want it.

At exactly this moment, a new kind of buyer has started calling. They call themselves AI rollups, AI-native holding companies, or permanent-capital owner-operators. The largest of them, Thrive Holdings, raised $2 billion at a $12 billion valuation in August 2026 to buy accounting, IT services, and regulatory businesses and run them on a shared AI layer. The pitch is polished and the check is real. I mapped 36 of these firms and graded every one on what it can prove, and I will come back to what that map says.

First, the part that matters more.

What AI actually changes about succession

Owners of businesses like yours have historically sold for two reasons that sound different and are really the same reason.

The first: nobody can run it. The next generation is not ready, or not interested, or there is one capable child and three siblings who want cash. The second: the business is you. The bids, the customer calls, the judgment about which crew goes to which site, the sense of which "urgent" means Tuesday and which means tonight. It lives in your head and in the heads of a few people who have been there fifteen years, and a buyer's diligence team cannot find it in the ERP because it was never there.

Both reasons come down to the same fact. The company runs on tacit knowledge held by too few people, and it takes too many other people to execute what those few decide.

That is the fact AI changes, and it changes it in your favor.

In the businesses we work inside, execution is getting radically cheaper. Drafting the proposal, reconciling the job costing, triaging the inbound, pulling the numbers for the Monday meeting: the routine layer of a services firm can now be done by a small team directing machines rather than a large team doing it by hand. The rollups are not shy about this. Fura, which buys freight brokerages, reports taking an acquired broker from a $150,000 loss to $1 million in profit with headcount falling from 26 to 8. Thrive's accounting platform reports its Tax AI drafting roughly 7,000 returns with about a third of preparer time saved. Those are the buyers' own numbers, unverified by anyone outside their cap tables. But take them at face value for a moment and notice what they imply.

They imply that the next generation, or one strong operator with a small bench, can run the business your parents built with a fraction of the headcount it took you. And they imply that the knowledge in your head can be moved, deliberately, into tools and playbooks the whole company uses, which is the exact work these buyers plan to do the week after closing.

So the two classic reasons to sell have weakened at the same time. "Nobody can run it" is less true when running it needs fewer people. "The business is me" is less true when your judgment can be captured and reused. What is left is a genuine choice, and a choice is worth a great deal more than a forced sale.

Now look at the rollup pitch again. The thesis they are paying a premium for is that AI can transform a business like yours. That is your opportunity. They are offering to buy it at today's multiple, do the work, and keep the difference. General Catalyst, which coined the phrase "AI-enabled roll-up," says as much: the enterprise-value uplift goes to whoever pairs capital with the transformation capability. You already have the business. The capability is available for fees, not equity.

The four real options

Every option ends at the same destination, a business that runs on an AI operating layer. They differ in what you give up to get there. I laid out the same four paths on the owner's section of the market map; here they are in succession terms.

1. Hand it down. The next generation takes the keys, and the equity moves by gift, sale, or a mix, on your timeline. The old objection was capability. The new answer is that a successor who can direct AI tools does not need your thirty years, and the tools can carry a great deal of what you know. What this path needs is a successor who has run a real outcome inside the business, with a number attached, before they run the whole thing.

2. Hire an operator and keep the equity. The family keeps ownership and a professional runs it. AI makes this cheaper and less risky than it was, because the operator runs a leaner company with more of the founder's judgment written down. The failure mode is hiring a manager for the business you have today instead of the business it could be. Fix the business first, then hire for it.

3. Sell, to a strategic or to a rollup. You get liquidity and the transformation is done to the business you built, by someone else, for their benefit. Sometimes that is exactly right: no successor, no appetite, a strong price. The permanent-capital buyers hold what they buy and usually leave selling partners with a rollover stake; Thrive says in its own words that it holds forever. The venture-backed platforms run the same playbook on a fund clock, with an expected exit and terms built around it. Whichever buyer, insist on the five questions below.

4. Hold and compound with an operating layer. Keep ownership, bring in the engineering capability the rollups use, pay for it in fees or against outcomes, and keep the margin it creates on your side of the table. This is the lane my firm works in, so read that with the appropriate discount. I name it because it exists, and most owners I meet have been told it does not.

None of these is right for everyone. What I am arguing against is the frame in which option three is the only one on the table, and the buyer is the only one who can do the work.

What the buyers can prove

Before you weight the sale option, know what the evidence says, because it is thinner than the coverage suggests.

On the AI Rollup Market Map I grade 36 firms on the quality of their public evidence: A for an outcome confirmed by a credible third party, B for a client-named or company-reported result, C for a real thesis with no verifiable outcome yet. As of September 2026, two firms hold grade A, and both are conventional private equity, Vista and Apollo. Every owner-operator, rollup platform, and transformation partner on the map, my own firm included, sits at B or C. Nobody in the new category has published an audited before-and-after on a business it owns.

The base rates are not kind either. Paul Carroll and Chunka Mui's study of 750 major business failures in Harvard Business Review found that "more than two-thirds of roll-ups have failed to create any value for investors", usually because the pace of acquisition outran the integration. A Fortune op-ed by two AI investors called the model, at best, tech-enabled private equity: services businesses trade at services multiples, and an AI layer any competitor can rent for a subscription does not change that. Equal Ventures adds that cost savings from AI are "largely ephemeral" because the firm next door will install the same tools.

I wrote the full argument, with the bull case given its due, in Do AI rollups work?. The short version for a seller: the buyer's model assumes a re-rating no one has yet earned, and you are being asked to price your life's work off that assumption.

Five questions to ask any buyer

Whether the firm across the table is a rollup, a venture platform, an outside operating partner, or mine, ask these before you sign anything.

1. Show me before-and-after numbers from two businesses you already own or serve, audited or at least client-confirmed: margin at close versus today, and how long the improvement took.
2. What does your headline AI number measure, exactly, and who outside your company checked it? "30 percent time saved" can mean a dozen things.
3. What happens to my team, my brand, and my client relationships in year one, and which of your prior acquisitions can I call to confirm it?
4. If I roll equity, what is the second bite worth today, at whose valuation mark, and when does it become cash? An accounting-industry newsletter put the warning well: an offer is often a multiple on cash flows that AI may compress, wrapped in an earn-out tied to margin targets that depend on an AI rollout you will not control.
5. If we part ways, who keeps the software, the data, and the people you trained?

The grades on the map are a head start on the first question. If a buyer cannot answer the other four in writing, you have learned what you needed to.

Keep your options open: fix the business first, then decide

Here is the sequence I would run, whether you end up handing the company down, hiring for it, or selling it.

Run one outcome inside your own company before you talk to anyone about price. Pick the process that eats the most skilled time. Write the success criteria down before the work starts, with a number and a date, and name a judge inside your company who decides whether it hit. Keep the software, the data, and the trained person whether it works or not. Eight to twelve weeks is enough to know.

Three things happen when you do this.

The valuation conversation changes. A buyer's AI thesis rests on the assumption that your margins can move. When you have already moved them, with a number a client or an auditor can confirm, you are selling proof instead of a promise, and you should be paid for it.

The succession conversation changes. The person who ran that outcome, whether that is your daughter, your operations lead, or an operator you hired for it, has just done the hardest part of the successor's job in public. You know something about them a title could never tell you. At a private-equity-backed industrial services client, the manager who owned the help desk was shipping his own code to production four weeks into the work, and nobody had to guess whether he could run the next one.

And your own conversation changes. Some owners run the first outcome and want another decade. Some run it and are ready to sell, on better terms. Either way the decision gets made with information instead of fatigue.

What I would avoid is the opposite sequence, which is the one the buyers prefer: sign a letter of intent, let their team do the transformation after closing, and take the second bite on their mark, on their timeline. It is a rational sequence for them. It is the expensive one for you.

One more door: if the goal is for the business to outlive the family's ownership, the people who work in it are natural long-term owners, and an employee stock ownership plan is the practical form of that. How that fits with permanent capital is a longer argument, and a later essay will make it.

Where to start

If you recognized your company in this essay, the first step is small, and it is the same whichever option you choose.

We call it Groundwork. It starts with a one-time $1,000 activation, then bills live time with your team and the compliance work your organization requires, by the minute, under a hard monthly cap set up front. The analysis, data mapping, prototypes, and write-ups in between are included and not time-billed. It is how we learn a business from the inside and find the first outcome worth running, with success criteria written down before any work starts and a judge you choose. Our public outcomes ledger lists the engagements resolved as successes and the success rate across every resolved engagement, failures counted, and we grade ourselves B on our own map because our results are client-named but vendor-published. We would rather you hold us to that standard than take our word.

Some clients, later, invite us to invest. Most do not, and the relationship stays where it started. Either way, you will make your succession decision with the business already fixed, and that is the whole point.