Advise. Build. Operate.

Most of the value in an AI project is decided before any code is written: which job it does, what data it needs, how you will know it works, and what it costs every month once it is live. We look at the work your team does, find the jobs where AI earns its keep, and price them honestly, including the ones we would advise against.

You get

• ›A ranked shortlist of AI opportunities in your business, each with a build cost and a monthly running cost
• ›A plan for the first one: scope, data, how success is measured
• ›Or, if you already have a quote from another firm, a line-by-line second opinion on it

Typical work

• AI readiness review for a team about to commission work
• Build-or-buy decision for a specific AI feature
• Review of an agency proposal before signing

We build the thing and connect it to where the work already happens: a support inbox that drafts replies from your own help pages, a document store you can search by meaning, an app with an assistant inside it, an intake process that reads and files what arrives. Each one ships with tests on your real cases, so you can see that it works before your team relies on it.

You get

• ›Working software in production, in repositories and accounts you own
• ›An evaluation set built from your own cases, rerun when models change
• ›Model usage billed at cost on its own line, or to your own provider account
• ›Notes your next developer can pick up without us

Typical work

• Customer-facing app with AI features
• Assistant over internal documents
• Automated intake, triage or drafting
• AI added to an app you already have

AI systems drift: providers change models, prices move, your data and your processes change. We watch the measures that matter for your system, rerun its tests when anything underneath it changes, and move it to a better or cheaper model when one passes the same tests. You get a short monthly note of what changed and why.

You get

• ›Monitoring of the measures that matter for your system
• ›Tests rerun when a model or provider changes
• ›A monthly note of what changed, what it cost and what we recommend

Typical work

• Ongoing care for a system we built
• Taking over a system another firm built