What to expect, from first conversation to running system.
Most firms we talk to haven't done this before. There's no standard playbook to compare us against, no procurement process that fits neatly. So here's exactly how an engagement works — what happens at each stage, what we need from you, and what you get at the end of it.
The first conversation.
It starts with a call or a meeting — no deck, no formal brief required. We want to understand what your firm does, where your team is currently with AI (whether that's "everyone uses ChatGPT informally" or "we've tried a few things and none of them fit"), and what the work looks like that you'd most want AI to help with.
By the end of that conversation, you'll know whether what we do is the right fit for you, and we'll know enough to say whether we think we can genuinely help. No commitment on either side.
Discovery — understanding the work properly.
If the first conversation suggests there's something real here, we run a short, fixed-scope discovery engagement. This is 1–2 weeks of structured work where we go deeper: we sit with the people who will actually use the system, we look at the document types and workflows involved, and we run a small technical test to validate the approach.
At the end you receive a written proposal: what we'd build, how long it would take, what it would cost, and what we're explicitly not doing. Everything scoped and priced before you commit to anything larger.
What we need from you: Access to a sample of your documents (under NDA), time with 2–3 people who understand the workflows, and a contact in your IT team if relevant.
What you get: A written proposal you can take to whoever needs to approve it. Even if you decide not to proceed, the scoping document is yours — it describes the problem and a viable approach to it, which has value regardless of who builds it.
The build.
Once the proposal is agreed, we build. This typically runs 4–12 weeks depending on scope — a single focused application at the shorter end, multiple integrated tools with system connections at the longer end.
While we build the applications, we're also procuring and configuring the server hardware and arranging colocation at a professional datacentre in your city. The two tracks run in parallel so they're ready at the same time.
How you stay involved: Weekly demos — you're reviewing working software, not mockups or status updates. You see the actual thing and give feedback in real time. The build changes based on what you tell us.
What you own: Everything. The code, the data, the model configuration. Nothing is locked to us — if you ever want to hand the system to someone else or manage it yourself, you can.
Going live.
When the applications are ready, we bring the production server live at the datacentre, migrate everything across, and connect your team. We train the people who'll use it — not a lengthy formal training session, but working through the actual tasks they'll use it for until it feels natural.
You also get 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 should be able to answer those questions independently.
From this point, your team is using a system that knows your firm's documents, responds in your language and format, and runs on hardware in your city.
Ongoing — managed, maintained, and improving.
After go-live, we manage the system on a fixed monthly basis. That covers security maintenance, monitoring, and model updates — when a meaningfully better open-source AI model is released, we update your deployment. Your team wakes up with a more capable system without any disruption or additional cost.
Most clients also keep us on a development retainer for ongoing work — new applications, additional integrations, and after 6–12 months of usage, fine-tuning the model on your firm's own data. That progression — start with your document library as context, then deepen with real examples over time — is how the system becomes genuinely irreplaceable rather than just useful.
Scale and scope vary. The process doesn't.
One application, one department
A focused first engagement — one well-defined tool for one team. Shorter build, lower hardware requirement, faster to value. Typically the right starting point for firms that want to prove the concept before committing further.
A suite of tools across the practice
Multiple applications — document review, drafting, billing capture — built together and sharing the same document library. The recommended starting configuration handles this comfortably for firms up to around 50 staff.
Multiple departments, higher throughput
A larger hardware configuration, dedicated server, and potentially different applications per department. The same process and the same managed service model — the scope and the hardware specification are what changes.
The first conversation costs nothing.
Two paragraphs about what your firm does and what you're hoping AI can help with is enough to start. We'll respond inside one business day and suggest a time to talk.