Pricing

What the one-time build includes

A single agent, shipped end-to-end done for you, tuned against your real data, with written handoff documentation. Specifically: a 20-minute scoping call, a written spec delivered within 24 hours, 3–5 days of build (spec dependent), 1–2 rounds of revisions after review, a handoff document, a Loom walkthrough, and 30 days of email support for bugs or small tweaks.

What it doesn’t include: ongoing prompt tuning after month one, new agents after the first, custom compliance work (NetSuite builds, HIPAA workflows, EU data residency, all quoted separately), or training your team on how to own the agent long-term. That last piece is included in ongoing support.

When ongoing support pays for itself

• You want three or more agents built in a two-month window, and ongoing support covers several per month at a better per-agent cost.
• You need continuous tuning. Agents drift as your data and processes change, and someone needs to notice and adjust.
• You’re standardizing on AI across teams and want a single external partner rather than a DIY-then-hire cycle.
• You want async Slack or email access for questions, edge cases, or ‘can you build a quick thing by Friday’ asks.

What about the AI provider’s own cost?

Some platforms are free during a research preview, after which the provider bills usage-based credits (model, runtime length, tool calls). Typical single-run agents (a weekly metrics reporter, an invoice reviewer) cost pennies per run. Conversational agents (lead outreach, support triage) run to a predictable monthly amount per rep at typical volume.

During build we size agents for predictable cost and set per-agent credit caps. You’ll know roughly what you’re signing up for before agents run in production.

Questions