From first conversation to running system.

Discovery and scoping

We sit down with you — in person where practical — and map the problem: which workflows matter most, which documents need to be indexed, which systems need to be integrated, and what a successful outcome looks like.

We run a small technical spike to validate the approach. At the end, you receive a written proposal: fixed scope, fixed price, explicit list of what we're not doing, and a timeline. No surprises after you sign.

$4,000–$9,000 depending on scope

Build

We build the applications against a development environment while simultaneously arranging hardware procurement and colocation. Weekly demos keep you informed; you're reviewing real working software, not mockups.

We work in your stack where it makes sense and propose alternatives where it doesn't. Every bill is itemised. Compute is billed at cost, transparently. You own the code, the data, the model weights, and the keys — not us.

$1,400–$2,200/day per engineer

Deploy and handover

We configure the production server at the colocation facility, migrate from development environment, set up VPN or network access for your team, run acceptance testing, and train your staff on the applications.

You receive written documentation of the system: what's running, where it's running, how to access it, and what to do if something goes wrong. Your team is capable of answering those questions independently.

Managed service

From go-live, the system is on a monthly managed service fee: security patches, model updates, monitoring, and support. We handle the infrastructure; you handle your work.

Clients who want ongoing development capacity add a retainer. Clients who don't can call on day-rate work for specific projects. Either way, no bills come as a surprise — every piece of work has a written estimate and a stop-work threshold.

$2,000–$8,000/month managed service

How long until something is live?

A typical first application — document Q&A over an existing document library — is live within 4–6 weeks of signing. More complex builds with system integrations take 8–12 weeks. Hardware lead times are 2–3 weeks, so the schedule usually isn't blocked on that.

Do you need to come to our office?

For the discovery stage, yes — we'd rather spend a day with the people who'll actually use the system than try to map their workflows from a video call. After that, most work is remote. We'll attend on-site when it adds value.

What do you need from us?

Access to a sample of the documents we'll be indexing, someone who understands the workflows we're automating, and IT contact information for any systems we need to integrate. We don't need a formal RFP or an elaborate brief — a conversation is enough to start.

What if the model improves significantly?

When a meaningfully better open-source model ships, we update your deployment. It's included in the managed service fee — no additional charge, no new project, no disruption. You benefit from the accelerating open-source frontier automatically.

What happens if you can't fix a problem?

We post a public post-mortem within 5 business days of any serious incident. SLA credits apply automatically if we miss 99.9% availability in a calendar month. We don't hide problems.

Can we cancel the managed service?

Yes. 30 days notice. You own the hardware, the code, and the data — cancelling the managed service doesn't take any of those away. If you want to manage the system yourself or hand it to someone else, you can.