The interview subject is the CEO of a Series D growth-stage company at roughly $410M ARR. She had not personally joined any vendor sales call in the eighteen months prior to her company's AI vendor evaluation. She joined the architecture call. We sat down with her for thirty-one minutes about why, what the call delivered, and what she has done since. She asked us not to name the company because the AI program is still in the install phase and procurement specifics are sensitive.
On why she joined.
Knyte: You had not joined a vendor sales call in eighteen months. What was different.
CEO: Two things. First, the AI procurement decision is going to be load-bearing for the next three years of the company. Most vendor decisions are not. Most vendor decisions are about specific tooling that we will replace inside two years if it does not work, with bounded switching costs. The AI procurement decision is about the architecture our institutional context is going to live on. The switching cost is not bounded. I joined because the consequence of getting this wrong is asymmetric.
CEO: Second, my procurement team had told me that this particular vendor's sales motion was different. The architecture call is not a demo. It is a working session with the vendor's architect that produces a written replacement-math sketch and a deployment plan. My head of procurement said it was the most useful first meeting she had taken in years. She thought I should see it.
On what the call delivered.
Knyte: What did the call actually produce.
CEO: It produced two documents and a decision. The first document was a workflow inventory of our current AI tooling — twelve workflows across five vendors that the architect had compiled from a fifteen-minute walkthrough with my procurement and engineering leads. The compilation was more accurate than the inventory my own team had been maintaining. That was a small thing that signaled a larger thing.
CEO: The second document was a replacement-math sketch — for each of the twelve workflows, a rough estimate of what the architecture install would replace, what it would cost, and what the compounding curve looked like. The sketch was honest. Three of the twelve workflows did not justify migration. Two were borderline. The other seven made a clear case. The honesty was the thing that made me trust the rest of the conversation.
CEO: The decision was that we would proceed with a paid scoping engagement, not an install. The architect explicitly said the install was not yet justified — we needed to validate two of the borderline workflows before we committed to the architecture investment. That recommendation was unusual enough to be diagnostic. The vendor was telling me to spend less, not more, in the next quarter. That is not the standard sales motion.
On what she has done since.
Knyte: What has happened in the months since.
CEO: The scoping engagement validated the two borderline workflows. We are now in week six of a ninety-day install covering the seven workflows the architecture call had clearly recommended. The compounding curve is bending exactly the way the sketch predicted. The two borderline workflows we ended up adding are also performing in line with the scoping engagement's projections.
CEO: What I have changed in my own behavior is more interesting. I now join the first call of every vendor evaluation that touches institutional context — anything where the asset we are building lives on the vendor's infrastructure. The bar is high enough that I only do this two or three times a quarter. The pattern is consistent: the first call tells me much more than the demo or the deck would have. I should have been doing this for years.
On the framework she uses now.
Knyte: What are you actually evaluating in those first calls.
CEO: Three things. First, can the vendor accurately describe my current state in the first thirty minutes. If they cannot — if the conversation is mostly them describing their product — they are not yet thinking about my deployment specifically. Second, does the vendor produce a written artifact in or immediately after the call. The artifact is evidence that the vendor's process is rigorous; the absence of one is evidence that the sales motion is the entire product. Third, would the vendor recommend against buying their product if the math did not support it. I now ask this question explicitly. The answers separate the vendors I will work with from the vendors I will not.
CEO: The third question is the one that filters most aggressively. Most vendors cannot give an honest answer because their compensation structure punishes honest answers. The vendors who can are the ones whose product economics align with the buyer's outcome economics. Those are the vendors I keep in the consideration set.
On what she would tell other CEOs.
Knyte: What is the message.
CEO: Two things. First, the AI procurement decision is the most consequential vendor decision your company will make in the next three years. The asymmetry of getting it wrong justifies your personal time at the first meeting, even if your procurement team is excellent. Your procurement team can evaluate the vendor's substance. Only you can decide whether the vendor's posture is one your company can build a multi-year program with.
CEO: Second, take the architecture frame seriously. The vendors who organize their sales motion around an architecture call rather than a demo are signaling something important about how they think about your deployment. The conversation is more useful, the artifacts are more durable, and the procurement decision is easier to defend. I do not think this generalizes to every vendor category. It generalizes to the AI category as I have experienced it, and the procurement decisions I have made under this frame are the ones I am most confident in.