AI Value
Fix the process first. Here is the math.
Two companies buy the same AI tool from the same vendor. One ends up with an asset. The other ends up with a liability. The difference was the process.
I put numbers on this so an investment committee or a board can argue with them. Here is the example I use.
How I estimate it
Net AI Value is the risk-adjusted net present value of an AI use case. In plain English: what it should save, after what it costs to build, run, and change, and after a reserve for what could go wrong.
It has two outputs, and I never add them together.
- Hold-period value: the present value of the cash the use case captures, minus the present value of what it costs to capture, minus a reserve for discrete loss events.
- Run-rate EBITDA at exit: the recurring cash benefit minus the recurring run cost, in the year a buyer will underwrite. A buyer pays a multiple of baselined, recurring EBITDA.
Two rules keep it honest. Redeployed time is capacity, not cash, so it does not count. Adoption shortfall lives in the adoption assumption, not in the risk reserve, so nothing is counted twice.
The example
Illustrative example. Hypothetical numbers and assumed inputs. Not a client and not a result. Your results will differ.
A $150M-revenue distribution business. 60 people in order-to-cash and customer service at $70K loaded cost ($4.2M). AI could free 25% of the effort ($1.05M a year of capacity). Assume 60% converts to cash through hires avoided as volume grows: $630K a year at full adoption. Year one runs at half speed while the change lands. Three years left in the hold. Discount rate 12% (20% shown as sensitivity). One-time build and change $400K; run cost $120K a year.
| Case | Assumed adoption | Extra cost | Risk reserve | Hold-period value at 12% (20%) | Run-rate EBITDA at exit |
|---|---|---|---|---|---|
| Sound process | 80% | none | $80K (data/IP $20K, security $30K, vendor exit $30K) | +$217K (+$119K) | +$384K a year |
| Broken process, automated as is | 35% | $100K a year rework | $170K (adds error at scale, 15% × $600K = $90K) | −$667K (−$661K) | about $0 |
| Fix the process first, then automate | 80% | $150K process fix; 6-month delay | $80K | −$45K (−$136K) | +$384K a year |
What it shows: automating the broken process loses about $670K and leaves nothing a buyer will pay for. Fixing the process first roughly breaks even in cash inside this hold and leaves $384K a year of recurring EBITDA that a buyer can underwrite at exit. Break-even adoption for the sound process is about 62% at 12%.
Net AI Value is a planning estimate, not a valuation. It is my structured estimate of the value an AI use case may capture, less the cost to capture it, less a reserve for identified risks. It is built from information the company provides and from assumptions that I label. I have not audited that information. It is not a valuation, fairness opinion, appraisal, forecast or guarantee of any result. It does not set or recommend a purchase price, and nobody should use it as the sole basis for an investment, pricing, financing or other decision. Actual results can differ materially. It is not legal, tax, accounting, regulatory or investment advice, and I am not a lawyer; items that touch legal or regulatory matters are flagged for your own counsel. It is prepared for the client named in the engagement letter, and no one else may rely on it. Illustrative examples on this site are hypothetical and do not describe any client or any result.
What the example says
Compare the broken case with the fixed one. Same tool, same vendor, same buyer. Fixing the process first costs $150K and six months. It turns a loss of about $670K into roughly break-even inside the hold, and it leaves recurring EBITDA a buyer can credit.
That is why I screen every AI use case before anyone estimates it. The Amplifier Test has five gates: value, process, data, people, and risk. In the broken case the process gate is red. Any red gate: do not fund it yet. The verdict is "Fix the process first," not "Fund." In my experience most AI ideas fail at least one gate.
Others have seen the same pattern. Google's DORA research found that AI acts as an amplifier of a team's existing strengths and weaknesses (DORA, State of AI-assisted Software Development, 2025). McKinsey found that AI high performers, about 6% of respondents, were nearly 3x as likely to have redesigned workflows (McKinsey, The state of AI in 2025, November 2025). That is a correlation, not proof of cause. I read it as a hint: the leaders fixed the work the tool was pointed at.
Why I give a range
I never hand over a single number. I show Low, Base, and High, the three assumptions that move the answer most, and the break-even adoption. Net AI Value is an estimate you can argue with. Every line is an assumption with an owner. If you disagree with my 35%, change it and watch the answer move.
What I would ask you
Pick one AI initiative you are funding. Answer three questions.
What named P&L line does it move, and what is the baseline today? Who owns the process it runs on, and what is the error rate? If it goes wrong, can you switch it off?
If you cannot answer them, you have found the cheapest money to save in your AI budget.
I show how I build Net AI Value on the AI value page. If you are an investor estimating the AI in a deal, start with the deal lifecycle.
Not ready for a call? Start with the free AI Leadership Readiness Assessment.