Before any vendor demo, there’s a cheaper and faster test than any proof of concept: three questions that reveal whether a company is ready for AI — not by the size of its budget, but by the structure that already exists (or doesn’t) behind the process it wants to automate.
The test isn’t about budget, it’s about structure
Most companies assess AI readiness by asking the wrong question: “do we have the money for this?” Budget solves buying the tool, not its outcome. What determines whether an AI project generates returns are three structural conditions, present or absent long before any vendor enters the room:
- Clear problem: is there a specific definition of the expected business outcome — cutting cost, growing revenue, shortening response time — or did the project start because “the competition is already using AI”?
- Defined process: is the routine AI will take over documented somewhere, with explicit rules, or does it only exist as tacit knowledge held by whoever has always done it that way?
- Organized data: is the information feeding that routine structured and accessible, or scattered across personal spreadsheets, emails, and an employee’s memory?
Missing any one of the three, the project isn’t ready for AI yet — it’s ready for an earlier, less visible, and more decisive step.
Why “process in someone’s head” isn’t a process
The second condition is the one most companies think they already meet — and the one that fails most often in practice. A process only exists, for AI purposes, when it’s formalized outside the head of whoever executes it. Unwritten rules, exceptions “everyone just knows,” and decision criteria that shift depending on who’s on duty aren’t material an AI agent can operate on.
João Paulo Batistella, an innovation executive and organizational transformation specialist, sums up why: the premise that used to justify how companies hired no longer holds — the question used to be “how do I get the right people to execute this”; now, with AI as a foundation, not just another tool, the question becomes “what am I going to ask this available brain to do.” But that brain only executes what’s formalized: a routine that lives only in one person’s memory has no basis for AI to act on it at all — formalizing it as data comes before automation, never after.
That explains why so many AI projects “don’t perform as promised” without the technology ever failing: the tool worked exactly as designed, but there was no formal process for it to follow — only the expectation that it would guess what nobody had written down.
The checklist before any vendor
Five questions, answered before the first meeting with an AI vendor:
- If the person who currently runs this process left the company tomorrow, would there be enough documentation for someone else to take over without rebuilding it from scratch?
- Does the expected outcome fit in one objective, measurable sentence — or does it only exist as “improve efficiency”?
- Is the data feeding this process in a system, or does it depend on someone copying and pasting from different sources every time?
- Is there a defined owner for deciding what AI can and can’t do in this process — or will that be figured out after something goes wrong?
- Does whoever is approving the investment know how to explain why this specific process, and not another, was chosen first?
Two or more uncertain answers indicate the missing work isn’t picking a vendor — it’s organizing the operation itself.
The limit of this test
This framework filters out obvious waste, but it doesn’t replace business judgment: some processes are worth formalizing even without any organized data today, because the expected value of the outcome justifies the upfront investment in documentation. The test doesn’t say “don’t do it” — it says “do the organizing step first, with eyes open to the real cost of skipping straight to the tool.”
Follow Eleva Tecnologia for more on technology applied to business: follow @ElevaTechno on X or @elevatechnologies on Instagram, or learn more about the group at elevatec.net/about.
