Essay

What Your AI Vendor Will Not Tell You

#cxo#procurement#strategy

Your AI vendor is not lying to you. They are just answering the questions you ask, and the important questions are the ones that do not come up in the demo. The demo is designed to impress. The contract, the cost curve, and the lock-in are where the real story lives. Here are five things worth asking before you sign, because the vendor will not raise them first.

Five things to ask about

  1. The cost at scale, not the cost of the pilot. The pilot is cheap by design. Ask for the fully loaded cost per outcome at your real volume, including the model calls a single user action triggers. Pilots are priced to get you in. Production is priced differently.
  2. What happens to your data. Ask plainly: is our data used to train your models, is it isolated, can we get it back, and does your platform become the only place our accumulated data and evals live. If your moat ends up inside their system, that is the real price.
  3. How hard it is to leave. Ask what it takes to switch models or vendors in eighteen months. If the architecture ties you to their weights, their format, or their proprietary layer, you are not buying a tool, you are buying a dependency, and they know it.
  4. Who owns the accuracy. When the model gets something wrong in production, whose problem is it. Vendors happily own the demo and quietly disown the edge cases. Ask where their responsibility ends and yours begins, before it matters.
  5. Where the price goes next. You do not control the per-unit price of the intelligence you are renting. Ask what has happened to their pricing, and what leverage you have if it moves against you. The margin you save today can be repriced away tomorrow.

The pattern

Every one of these is about the same thing: who captures the value in three years. The vendor's honest goal is to make themselves indispensable and your switching costs high. That is not villainy, it is strategy, and your job is to negotiate against it with open eyes. Own your data, keep the model swappable, price on outcomes, and read the exit clause as carefully as the feature list.

So before your next AI contract: which of these five did the vendor not bring up, and why do you think that is?