Three of our recent calls were with technology leaders who were already spending on AI infrastructure. One booked a meeting. The other two told us they had something in place or weren’t ready yet. The difference between those outcomes came down to what we learned in the first minute.

The booked call was with a VP of Engineering. Within the first few minutes, he described the problem himself: he was struggling to govern and consolidate costs for AI agents spread across Microsoft, Anthropic and GitHub. We didn’t have to explain the pain. He agreed to an intro call, gave us his email, and said he would bring an engineering manager. The call ran 936 seconds.

The second call was with a lead at a small engineering firm already running AWS Bedrock and Google APIs. He could see the value of a gateway for cost control and security once they scaled, but said they were too small for it today. That is a timing answer, and it arrived early. The right response was to accept it and set a dated follow-up.

The third was a director of business development at a firmware company that runs an internal SDR team. She was skeptical of outsourced agencies after past bad experiences, and she doubted an agency could speak credibly about a technical product. The rep addressed her concerns on the call. It ended in a follow-up with a decision maker, but only because the objection was handled directly.

Call length told us nothing about the outcome. The booked call was the shortest of the three at 936 seconds, and the other two ran 944 and 996. What mattered was whether the prospect named an active problem with an owner.

So we now ask three things before any pitch. What are you already running for this? Is the problem active enough to have an owner and a budget? Has a vendor let you down on it before? If the answer is a working internal build or a live RFP, you are in a vendor evaluation, so ask for the timeline and who is on it. If the answer is “too small” or “not yet,” agree a dated follow-up and end the call cleanly.

One booked meeting from three calls is a small sample, and we’d rather say that plainly than round it up. The lesson still holds. Qualify the tooling first, confirm the pain second, and pitch last.

Notes (not part of the post):

  • The title’s numbers aren’t in the records you sent. The data covers 5 calls with 1 meeting booked, not 11 compliance calls with 4 booked and 4 declined for existing tooling. I didn’t use the 4/11 figures. The title above is replaced for that reason. If you have the full 11-call data, send it and I’ll rewrite around it.
  • The records aren’t regulatory change calls. The three I used are about AI gateway and cost governance. If this post is meant to be about the compliance campaign, those calls aren’t in what you sent.
  • Excluded two records. The home refinancing call isn’t ICP and has nothing to do with compliance. The retired CIO’s record is marked not relevant, and its summary is cut off.
  • Format. I left off the title to match “body text only.” Nothing has been published.