We made five cold calls this week to engineering and operations leaders. Four booked a meeting. The one that didn’t shows where a pay-per-meeting offer stops working.

The calls ran about 10 to 13 minutes each. Here is what separated the bookings from the rejection.

The strongest opener came from a staff engineer whose team already runs OpenAI and Anthropic models. The problem was not adoption. It was visibility. Nobody could say what the team spent on tokens or which workloads drove the bill. That is a trigger you can state in one sentence and confirm with one question. The call ended with a 20-minute intro on the calendar for Thursday. The first email bounced, so the calendar invite and follow-up go out by text or phone.

The second booking was a CTO at a company with 70 staff and 50 engineers. They opened by saying AI spend was already under control through their internal developer portal. Most reps would have ended the call there. We kept going, and the CTO got interested in the AI gateway as something to offer their large enterprise clients, not as a fix for their own team. A “we’re covered” answer can point you to a second use case. That meeting is set for Wednesday at 5pm.

The rejection came from the managing partner of an investment group that runs about 100 entities through a corporate services provider at around $200 per entity. Their process works, and speed is not a priority. Two objections came up: the disruption of switching providers, and whether a startup will still be around to support them. Neither is a price objection, so a lower cost per meeting does nothing for either.

Trigger points we now target:

  • Teams already paying for AI APIs with no spend breakdown.
  • Engineering leaders who see a path to resell or refer AI tooling to their own large clients.
  • Operators who say their process is fine. Deprioritise these unless you have an offer that addresses switching risk.

On the “agencies burned us” objection, we don’t defend the category. We ask what went wrong last time: volume without meetings, wrong titles, or no visibility into who was contacted. Then we show the difference in the first step, not on a slide. Define what counts as a qualified meeting in writing before the first dial. If the first step needs no integration and no change to their process, say so up front, because that is what answers the switching fear.

Four of five is a good week. The lesson is in the fifth. Target teams with a visible problem, and treat “we’re fine” as the start of a question, not the end of the call.