One vendor in 30 asked how the product went. It cost $50

SaaStr signed up for somewhere north of 30 APIs this year: search, enrichment, structured data, email infrastructure, inference, storage. The company runs 21-plus AI agents in production and is building a product called SaaStr AI Connect on top of them.

One of those 30-plus vendors asked how the product was working. Not one that asked better than the others. One, total.

The email came from Alina on Exa's product team, four days after Lemkin ran his first real batch of calls. It was four sentences long and offered $50 of credits for a one-line answer.

He replied 33 minutes later with roughly 400 words: a benchmark figure, a feature request, and his throughput and pricing constraints. She came back in under three hours with two endpoint configurations and a straight admission of what Exa does not yet support.

The mechanics are worth copying because none of them cost anything.

The trigger was usage, not signup. The email referenced testing, not account creation. Most onboarding drips fire on day 0, 1, 3 and 7 regardless of whether anyone made a call, which tells the customer nobody is looking.

The ask was explicitly small. "A one-liner" is a 15-second commitment; "do you have 30 minutes" is a calendar negotiation with a busy person who does not yet know whether you are worth it. The small ask is what produced the long answer.

The sender was a named human on the product team, from a personal address, with no template chrome and no unsubscribe footer stapled to a one-line note. Automated outreach that looks personalised is now infinite and cheap, which makes a real person with roadmap influence scarcer, not cheaper.

The incentive was denominated in the vendor's own product. Credits cost Exa almost nothing at the margin, so the motion runs at volume without a budget argument, and they buy more usage of the thing you want used.

The reply contained specifics, including a clear no. Lemkin's view is that "we don't expose that yet", from someone who understands the request, builds more trust than an enthusiastic maybe from someone who does not.

What Exa got back was a design partner conversation: a production use case matching roughly 145,000 B2B and AI executives, enrichment across about 3,000 companies, and a benchmark of 89% clean person matching on the adversarial pool其 structured-data vendor had returned not_found on. Also a named constraint — 10 queries per second — and the fact that the alternative tested at roughly five times the cost.

The rest of the field did nothing, ran a drip, or sent an SDR to book a demo of software already in production. One rep from Coresignal did make contact days after a tier upgrade, but the trigger was the credit card and the purpose was an upsell.

The arithmetic for a founder is simple. One product manager can send around 20 of these a week in a couple of hours; a quarter of that is roughly 250 conversations with people who have just used the product, plus a backlog written by paying customers.

The window is narrow. After a real test and before a decision, the opinion is formed and nothing is locked in. A week later the customer has either built on it or moved on, and the answer is worth less either way.