From first call to production in weeks, not quarters.
Diagnose in week one, build on your stack with weekly demos, launch against a baseline, then compound. Every step, what you get and what ends it.
One senior team owns the outcome end to end. Here is every step of an engagement: what happens, what you get, what we need from you, and the exit criteria that end it.
01 · Diagnose (Week 1)
We find the money before we spend yours.
Most growth problems are not a channel problem. They are a leak somewhere between the first click and the closed deal. Week one is about finding it, sizing it and deciding — together — what is worth building first.
What happens
- Kick-off with the people who own the number — revenue, operations or support — not only the person who signed.
- Funnel and data audit: where demand comes from, where it leaks, and what your stack already captures.
- Stack review across CRM, ad accounts, analytics, support tooling and data quality.
- Every opportunity scored on revenue impact, effort, risk and time to value.
What you get
- A ranked opportunity map (top three to five)
- A baseline for the metrics we will move
- A 90-day plan with go / no-go criteria
- Architecture notes any team could execute
What we need from you
- Read access to analytics, CRM and ad accounts
- Two to three hours of your team across the week
- One decision-maker who can say yes
Done when
You sign off the first system to build and the metric it has to move.
02 · Build (2–8 weeks)
Senior engineers. Your stack. Weekly proof.
No hand-off to juniors and no black box. The people who scoped the work build it, on your accounts and your infrastructure, and you see working software every week.
What happens
- Weekly demos of working software — never status slides.
- Built inside your existing stack and accounts. You own the code, prompts and data.
- Evaluations written before the build, so "good" is defined on your real tasks.
- Guardrails, human escalation and a rollback path designed in from day one.
What you get
- A production-ready system
- An evaluation suite on your data
- Runbook and documentation
- Access, cost and latency controls
What we need from you
- A technical contact on your side
- Staging or sandbox access
- Fast feedback on each weekly demo
Done when
The system passes its evaluation suite on your data and your team signs off the final demo.
03 · Launch (Go-live)
Released against a baseline, not a hope.
Launch is a measured release, not a press day. We roll out in stages, compare against the baseline from week one, and keep people in the loop wherever judgement matters.
What happens
- Staged rollout — shadow mode or a small share of traffic before full release.
- Measured against the week-one baseline on dashboards your team can read.
- Human in the loop for approvals, escalations and edge cases.
- Enablement, so your team knows how it works and how to override it.
What you get
- The live system
- A launch dashboard
- A team enablement session
- Incident and rollback plan
What we need from you
- An agreed launch window
- Stakeholders available for sign-off
- Channel owners for ads, sales or support
Done when
The target metric moves against baseline and holds through the first full cycle.
04 · Compound (Ongoing)
The second month should beat the first.
Most AI projects decay after launch. Ours are run, measured and improved — and every next system is built on the foundations of the last, so cost per result keeps falling.
What happens
- Monthly reviews of the number, not the activity.
- Prompts, models, routing and spend tuned as better options ship.
- The next ranked opportunity built on the same foundations.
- Continuous evaluation catches regressions before your customers do.
What you get
- A monthly performance report
- A prioritised backlog
- Regression-tested updates
- A falling cost per result
What we need from you
- A 45-minute monthly review
- Access kept current
- A say in what ships next
Done when
When you decide to take it in-house — with playbooks written to survive our departure.
Four rules we never break.
The people who scope it build it.
No hand-off to juniors and no black box. The senior team in the first call is the team that ships.
You own everything.
Built inside your existing stack and accounts. The code, prompts and data are yours from day one.
Judged against a baseline.
We set the baseline in week one and measure every release against it, on dashboards your team can read.
Working software, every week.
Weekly demos of real software on your data — never status slides.
Start with week one.
Book a 20-minute call, or scope the project in five minutes and get a written brief back.
Book a 20-minute AI Opportunity Call · Get a free AI opportunity map