Our process
Four phases. You can stop after any of them.
Nobody should commit a large budget to an AI system they have not seen work on their own data. So we built the process so you never have to.
- 011–2 weeks
Audit
We sit with the people doing the work and map what actually happens — not what the org chart says happens.
Every candidate process gets scored on volume, effort and how measurable the outcome is. You get a costed shortlist, and an honest note on anything that shouldn't be automated at all.
You get: A prioritised, costed shortlist you own outright.
- 022–6 weeks
Pilot
We build the single highest-scoring process, on your real data, with your real documents and edge cases.
You see output before anyone signs off a bigger budget. If the accuracy target isn't met, we say so rather than shipping something that quietly fails.
You get: One working system, measured against an agreed accuracy target.
- 031–3 weeks
Deploy
Integration into the systems you already run — CRM, Tally, WhatsApp, your database. Staff training, escalation rules, a rollback plan.
We deploy alongside your existing process first, not instead of it, so nothing breaks while confidence builds.
You get: Live, integrated, with your team trained on it.
- 04Ongoing
Measure & widen
We agreed a number at the start. Now we check it. If it moved, we roll out further; if it didn't, we stop and tell you why.
Under a maintenance retainer we also handle model drift, provider deprecations and cost monitoring.
You get: A before-and-after on the metric that mattered.
Engagement models
Four ways to work with us
Fixed-price project
Defined scope, milestone billing, agreed acceptance criteria. Best when you know what you want built.
Dedicated team
One or more engineers working only on your roadmap, billed monthly. Best for continuous product work.
Audit and advisory
Discovery, roadmap and technical review with no obligation to build with us afterwards.
Maintenance retainer
Monitoring, quality evaluation, cost tracking and model upgrades for systems already in production.
FAQ
Before you get in touch
How quickly can we see something working?
For a focused chatbot or extraction pipeline, two to four weeks from kickoff. Voice agents and vision systems take longer because they need tuning against your real environment.
What do you need from us?
Access to the people who do the work, real sample data, and one person on your side who owns the outcome. Projects without that third thing are the ones that stall.
How do you bill?
Milestone-based for projects, monthly for retainers. Model usage, telephony and WhatsApp conversation charges are third-party costs we model for you upfront and bill at cost.
What if the pilot doesn't hit the target?
We tell you, we explain why, and you decide whether to continue. We would rather lose the follow-on work than deploy something that quietly underperforms.
Do we own what you build?
Yes — full source, repository and deployment access transfer to you on final payment.
Next step
Let's work out what's worth building.
A 30-minute call. We'll tell you what we'd do, roughly what it costs, and whether it's worth doing at all.