Two plans
| Plan | Price | Capacity | Cadence | Response |
|---|---|---|---|---|
| Half | $8,500 a month | 10 build days | Weekly release, fortnightly roadmap call | Next business day |
| Full | $16,000 a month | 20 build days | Weekly release, weekly roadmap call | Same business day, plus on-call for production incidents |
Both plans include hosting and monitoring operations, dependency updates and model cost reporting.
Why AI software needs a retainer more than most
Software with an AI part keeps changing after launch even when nobody touches it. Model providers release new versions and retire old ones. Your cases drift: new suppliers, new document layouts, new questions from users. The evaluation set catches the drift, and someone has to act on it. A retainer pays for that work at a known monthly price, instead of a new project each time.
What a month looks like
Each week ends with a release. The roadmap call sets the order of work for the next one or two weeks. Ideas that came up during the sprint, which went on a visible “next” list, are the first items. Model cost is reported every month, so a jump in usage is seen before the invoice arrives.
How it ends
The first term is three months. After that the retainer runs month to month and ends with 30 days of notice. Everything stays in your accounts and repository, so ending it needs no handover.
How this applies to us
We take at most three engagements at once, so a retainer is a reserved slot. The retainer is offered at the end of a sprint, priced on the plan that fits the roadmap you have.