The "we paid for it once" promise — what it costs us, what it buys you

Grow2.ai signs every pilot contract with a clause: on Day 14, we hand over the source, the prompts, the integration credentials, and a runbook. The client could fire us on Day 15 and keep the agent running. Three of fourteen clients have done a version of this — two for budget reasons, one for political reasons. None of them have come back to bite us. This note is why the clause exists, what it costs Grow2.ai operationally, and why we plan to keep eating that cost.

What's in the hand-off package

• The repo — TypeScript service, deployment manifests, infrastructure-as-code (Terraform or Pulumi), CI config.
• The prompts — system prompts, tone profiles, refusal contracts, eval sets — all in version control.
• The credentials — every API key, every webhook secret, every integration token, in a structured handover document.
• The runbook — what to do when the agent breaks, who to call at each provider, what the alerts mean.
• The eval suite — 50–280 graded conversations the client can re-run before any prompt change.

What it costs Grow2.ai operationally

Three things, in order of cost.

1. Engineering discipline. Every shortcut Grow2.ai might be tempted to take — undocumented config, a hard-coded credential, a quick patch to production — has to be undone before Day 14. This shows up as ~10% engineering time on the pilot, which is real.
2. Lower retention rate. A client who has the source can leave. Some do. Our retention rate is ~78% at 12 months — agency benchmark for similar work runs ~65%, so we're ahead, but a vendor lock-in clause would push us higher.
3. Sales cycle complexity. Procurement teams sometimes don't know what to do with a vendor that volunteers source ownership. We've had three pilots delayed by weeks while legal reviewed the clause.

What it buys the client

The thing the clause does that nothing else does: it makes Grow2.ai stop selling and start serving. We can't coast on switching cost. We have to be the right vendor on month 13 the way we were on month 1. That changes how we run the practice — what we ship, who we hire, which clients we take.

Why we keep eating the cost

Two reasons. First, the studio's self-interest: we want to do this work for ten more years, and the way to do that is to be the kind of shop people recommend to peers. The hand-off clause is the loudest possible signal that we're aligned with the client. Second, the studio's self-respect: we've seen the alternative — agencies that hostage the source code and bill against renewal anxiety — and we don't want to be that. The clause is partly a contract with the client, partly a contract with ourselves.