The market spent the past year deciding what your business is worth. It also quietly started deciding what your software is worth, and those two verdicts are about to land on the same diligence checklist.
Two reports this month made the case from different directions. First, the sorting. BizBuySell's Q3 Insight Report showed businesses selling for more than $1 million averaging 3.73 times cash flow, against 2.30 times for those below that line. That is more than a full turn of multiple. BizBuySell credits much of the gap to financing: smaller deals lean on SBA loans, and larger ones attract buyers who don't need them. Financing follows evidence, though. Lenders and buyers both pay more for what they can verify.
Then the software. A Censuswide survey of 1,001 U.S. IT decision-makers, commissioned by Workato and reported by CIO Dive, found almost two-thirds expecting an AI agent in their own environment to cause a public security or compliance incident within the next 18 months. Workato sells agent-governance software, so weigh that accordingly. The direction still matters.
Put the two together and the lesson is simple. Buyers price risk, and in 2026 the risk lives in the tooling as much as in the books.
The market already sorted you
Same quarter, same economy, same buyer pool, and more than a full turn of multiple separates the two groups. Some of that is scale and access to capital. But the businesses that clear the line tend to share something an owner can control: clean records, documented add-backs, and operations that keep running when the owner looks away.
Manufacturing owners should take note. BizBuySell reported manufacturing median sale prices up 58 percent year over year on cash flow up 38 percent, with a 3.02 times multiple, the highest among the major sectors. Buyers are paying up for businesses that hold up under examination.
What the buyer's team is starting to open
Diligence still opens with the financials: three years of returns, a quality-of-earnings report, an AR aging. On larger deals, buyers' counsel have started adding AI to the request list: what tools the company runs, what data goes into them, who owns what comes out, and what the vendor contracts allow. In the lower middle market it is still uneven, and many buyers ask nothing yet. That makes this the cheap moment to get ahead of it.
An owner who cannot say what their tools do is asking a buyer to price a black box, and buyers price black boxes conservatively. An unwatched digital employee with access to customer email is a liability with a monthly invoice.
Use it, and write the handbook
None of this is an argument for keeping AI out of your business. Your employees carry risk too, and nobody's answer to that is to stop hiring. You manage that risk in two ways. You train people before they take on the work, and nobody runs the quoting desk on day one. And you write an employee handbook that sets out how people are expected to work, what they may and may not do, and what happens when something goes wrong. Digital employees deserve the same treatment. Each one gets trained and tested on its job before it goes live, and the business gets a short AI use policy that says which tools are approved, what each one is for, what data it may touch, where a person has to check the work, and what the tools are never allowed to do on their own.
A buyer looking at a well-run shop that uses AI under a written policy sees capability: more output from the same headcount, with the controls already in place. The same tools with no policy look like exposure. The difference between those two readings is a few pages of writing.
The register entry
The handbook sets the rules for everyone. The register applies them tool by tool, and it is smaller than it sounds. Every AI tool in the business, whether you call it a digital employee, an agent, an assistant, or an automation, gets a register entry before the job is done rather than after. Mine all have one. Each entry covers six things:
Plan on about twenty minutes per tool, updated when the tool changes. It is the AI version of the documented processes this newsletter has argued for all fall: transferability, applied to software.
Inside the business, the register turns your tools into supervised staff instead of an unwatched utility bill. At a sale, it answers the buyer's questions before their team thinks to ask, which is what good documentation has always done.
The twelve questions
These are the twelve I would want a client to answer before a buyer's team starts asking. Few buyers in our market will ask all of them today, but an owner with written answers looks like an owner who runs a tight business.
1. Can you produce a list of every AI tool, agent, and automation in use, with a named owner for each?
2. Who approved each tool, and is that approval written down?
3. Do you have a written AI use policy, and do you know whether employees are putting company or customer data into AI tools on personal accounts?
4. Which customer, financial, or employee data can each tool access?
5. Where does your data go when a tool uses it, and do the vendor contracts say how it can be stored or used for training?
6. Do any customer contracts, NDAs, or privacy laws restrict putting that customer's data into third-party AI tools, and are you within those limits?
7. Who owns what the tools produce, such as quotes, drawings, code, or marketing copy, and do the vendor terms confirm it?
8. Who checks each tool's work, and how often? 9. Do any tools send quotes, pricing, or other decisions to customers without a person reviewing them first?
10. Has any tool produced an error that reached a customer, and how was it handled and recorded?
11. Are the accounts, subscriptions, and credentials held in the company's name, and will they transfer at closing?
12. If you stepped away tomorrow, who would know how to supervise each tool?
Questions 3, 6, and 7 are the ones most likely to surprise you, because the answers usually live with employees and vendors rather than with the owner. Most of the rest are one register entry away, which is the encouraging part.
I put all twelve on a one-page checklist with space to record your answers, so you can hand it to whoever manages your tools. Check the comments below to get a copy.
Steve Duke spent about 30 years learning both sides of this problem. In his corporate career, he worked for Fortune 100 aerospace companies, including GE Aerospace, Lockheed Martin, and General Dynamics. At one of these companies, he ran a multimillion-dollar business unit with full P&L responsibility. He then left to run his own company and discovered that decades of training don't protect you from becoming the business yourself. A serious health event ended it. He exited, recovered, and built Lucensys™ Group around one idea: every owner exits eventually, and the only question that matters is whether they built something that can fund what comes next.
He wrote Failure to Exit about why most owners arrive unprepared.
His next book carries an old warning for owners: the best time to plant the tree was 20 years ago. The second best time is now.
Lucensys™ Group works with owner-operated manufacturers, distributors, and industrial-services companies doing $2M–$50M — reducing owner dependence and the risk discounts that cost owners at the table.
Charlotte, NC · lucensys.io · steve@lucensys.io · (704) 953-5608