Insights
Exit readinessNovember 19, 20266 min read

AI exit readiness: what buyers will ask, and what to have ready.

Buyers now test AI claims the way they test revenue. The evidence to prepare, the red flags to fix first, and a timeline that leaves room to do it properly.

Buyers now ask every company about AI, and the questions have moved on from “do you use it?” to “show me what it does to the numbers, and show me it is safe.” AI exit readiness means being able to answer both with evidence a buyer’s diligence team can check, instead of claims they will discount.

The work is mostly about proof, not technology. A company with two modest use cases that are live, measured and governed will tell a better AI story at exit than one with a dozen pilots and a slide.

The four things a buyer will test.

  1. Value. Which AI use cases are live, what volume they handle, and what they have done to cost, capacity or revenue. Expect the buyer to ask for the baseline and to tie the figures to the finance system.
  2. Durability. Whether the benefit survives the sale: who owns each system, whether it depends on one person or one vendor, and what happens when the underlying model changes.
  3. Risk. What data goes where, which vendors process it, what contracts and terms cover that, where humans approve outputs, and what has gone wrong so far.
  4. Exposure. Whether AI threatens the business itself: competitors or customers doing the work in-house with AI, pricing models under pressure, or a product that a general-purpose tool now replicates.

The evidence pack to prepare.

Build this as a living folder, not a last-minute data-room scramble. For each live use case:

  • One-page summary. The workflow, the owner, the date it went live and the volume it handles.
  • Baseline and current numbers. Adoption, hours saved net of review time, quality against baseline, running cost.
  • Benefit by type. Capacity, cost removed and headcount reported separately, with payback.
  • Architecture note. Systems touched, models and vendors used, where data is stored, and how a model change is tested.
  • Controls. Human approval points, evals, access controls, incident log and rollback plan.
  • Contracts. Vendor terms on data use and training, and any customer commitments about AI.

Add a short company-level AI policy, an inventory of AI tools in use (including the ones staff adopted on their own), and the roadmap of what is planned next with honest status.

Red flags that cost money at exit.

  • AI benefits quoted in the CIM that nobody can trace to a baseline or the P&L.
  • Customer data sent to tools under consumer terms, or with no record of where it went.
  • A key AI system built by one contractor, with no documentation and no internal owner.
  • Pilots described as products.
  • No answer to “what would a competitor with the same AI tools do to your margin?”

Each of these is fixable, but rarely in the three months before a process starts.

A timeline that works.

  • 18 to 24 months out. Assess where the company stands, pick one or two use cases with a clear path to the P&L, take baselines, and put basic governance in place.
  • 12 months out. Use cases live with real volume and tracked monthly. Tool inventory and data map complete. Vendor terms reviewed.
  • 6 months out. Evidence pack assembled and checked against the finance system. AI section of the equity story drafted, with every claim backed by the pack.
  • In process. A named person ready to walk the buyer’s technical diligence team through each system and its numbers.

Writing the AI section of the equity story.

Keep it short and checkable. State what is live, what it has delivered, how it is governed, and what comes next with the next buyer’s capital. Separate delivered value from future opportunity, and never blend them into one figure. A buyer who finds one inflated number will discount all the others.

Common questions.

How early should a portfolio company start on AI exit readiness?
Ideally 18 to 24 months before a planned process. Use cases need time to go live and build a track record that a buyer will believe.
Does a company need AI to sell well?
Not always, but it needs a credible answer on both value and exposure. A clear view of how AI affects the business, even if the answer is “modestly”, is better than silence or hype.
What is the most common AI diligence finding?
Claims that cannot be traced to a baseline or the P&L, followed closely by unmanaged data flowing into AI tools that staff adopted on their own.

This is the mirror image of buy-side work. For the questions a deal team asks before close, see AI due diligence for private equity. For getting use cases live and measured in time, the AI Value Sprint is the usual starting point.