Skip to content

Our process

Five phases. A stop point at every boundary.

You can end the engagement at any phase and keep everything produced up to that point. We structure it that way on purpose — it keeps us honest about whether the next phase is worth funding.

Working software every two weeks, not status reports

A typical engagement

Phase 03
2 weeksAlways deployable

Between demonstrations of running software

Phase 01 · Understand1–3 weeks
Phase 02 · Decide1 week
Phase 03 · Build4–20 weeks
Phase 04 · Prove2–12 weeks
Phase 05 · OperateOngoing

6–10h

Your time in discovery

45 min

Demo every fortnight

90d

Benefit verification

The five phases

What happens, when, and what you get at the end of it.

Including the part most proposals leave out: how much of your team’s time this actually costs.

Phase 01

Understand

We sit with the people doing the work and record what actually happens, including the exceptions nobody documented. Then we measure it.

1–3 weeks

What you receive

  • Systems and data map
  • Measured baseline: cost, cycle time, error rate
  • Constraint and risk register
  • Ranked opportunity shortlist

Your time: Roughly 6–10 hours across the phase, mostly from operators rather than executives.

Phase 02

Decide

A costed recommendation with the trade-offs written down — including the option of doing nothing, and the option of buying instead of building.

1 week

What you receive

  • Solution architecture and decision records
  • Fixed-price proposal where scope allows
  • Guardrails, permissions and escalation design
  • Success metrics agreed in writing

Your time: One two-hour decision session. You get the document three days before it.

Phase 03

Build

Two-week increments, always deployable, demonstrated in a running system. Evaluation and security review happen every increment, not at the end.

4–20 weeks

What you receive

  • Working software in your repositories
  • Automated tests and CI/CD pipeline
  • Infrastructure as code
  • Evaluation suite and observability

Your time: A 45-minute demo every two weeks. Async access to the team throughout.

Phase 04

Prove

Shadow or parallel running until the numbers agree with the baseline. This is where the exceptions that would have become incidents show up instead.

2–12 weeks

What you receive

  • Parallel-run reconciliation report
  • Tuned thresholds and exception handling
  • Rollback procedure, tested
  • Go-live readiness review

Your time: Operator review of exception cases — typically 2–4 hours a week.

Phase 05

Operate

Staged rollout, monitoring, and a genuine handover. We are finished when your team can run and change this without us.

Ongoing or handover

What you receive

  • Runbooks and architecture documentation
  • Dashboards and alerting
  • Team training sessions
  • 90-day benefit verification

Your time: Two training sessions, then whatever your team wants.

Engagement principles

The rules we hold ourselves to when it gets difficult.

Senior people, start to finish

The engineers who scope your work build it. No handoff to a delivery team you have not met.

Baselines before builds

We measure the current state first. Improvement should be a fact you can point at, not an impression.

Working software over status reports

Every two weeks you see the system running, not a slide describing it. Progress is demonstrated, not asserted.

Change priced in the open

Scope changes get estimated and quoted before they enter the plan. Nothing is silently absorbed into a drifting number.

Security reviewed continuously

Threat modelling at design review and security testing every increment — not a gate discovered two weeks before launch.

Handover is a deliverable

Documentation, runbooks and training are scoped, estimated and delivered like any other feature.

Ninety minutes · no charge

What the first session actually looks like.

No slides from us. Four to six people from your side, ideally including someone who does the work rather than only people who manage it.

  • You describe the process that keeps breaking, in whatever detail you have
  • We ask for whatever numbers already exist — volumes, handling time, error rates
  • We name the constraint you believe is immovable and test whether it is
  • You leave with a rough systems map and an honest read on whether this is worth pursuing

What to bring

  • The process, not the solutionDescribe what happens today. We will work out whether AI is even the right answer.
  • Whatever numbers you haveEven rough ones. A wrong baseline is more useful than no baseline.
  • The person who does the workThey know the exceptions. Their manager usually knows the version in the process document.
  • Your hard constraintsData residency, regulator, a system you cannot touch. Tell us early so we design to it.

Process questions

How this works in practice

Still deciding?

Send the question to a senior engineer instead of a form. You will get a straight answer, and a no if that is the honest one.

Ninety minutes, usually four to six people from your side, no slides from ours. We want the process that keeps breaking, whatever numbers you already have, and the constraint you believe is immovable. You leave with a rough systems map and an honest read on whether this is worth pursuing. No charge, no obligation.

Less than you expect and more than zero. Discovery is the heaviest — roughly six to ten hours spread across two to three weeks, and mostly from the people who do the work rather than from executives. During the build phase it is a 45-minute demo every fortnight plus async questions.

You can, at any phase boundary, and you keep everything produced up to that point — including code, documentation and the assessment findings. We deliberately structure engagements in stages so stopping is a real option rather than a negotiation.

Two-week increments with a demonstrable increment at the end of each. We are unceremonious about the ritual — no story-point theatre — but strict about the discipline: always deployable, always demonstrated, always reviewed.

We write the trade-off down as a decision record with the options, the risks and our recommendation, then you decide. If you choose differently from our recommendation we record that too and build what you asked for. What we will not do is quietly build our preference.

We come back and measure. The business case gets re-tested against the actual numbers and we publish the result — including when it came in under what we projected. That review is part of every engagement and it is not optional on our side.

Start the conversation

Ninety minutes. No slides. No obligation.

Bring the process that keeps breaking and whatever numbers you have. You leave with a systems map and an honest read — including if that read is that you do not need us.

What to expect

  • No pitch deck, no obligation
  • Senior engineers in the room
  • A written plan within five days