APPROACH
Five phases, no black box
Most software projects go wrong before anyone writes code — in the gap between what was asked for and what the work actually needs. This is how we close it, and what you get at every step.
Map
A week inside the work, not in a workshop.
We sit with the people doing the job and follow the process end to end — including the spreadsheet nobody mentions and the four-step workaround everyone has memorised. Requirements written from a wishlist describe the software people think they want; requirements written from observation describe the software they will actually use.
- Shadow the day-to-day, two or three roles deep
- Map every handoff, tool and duplicate entry
- Put hours and money against each friction point
- DURATION
- 1 week
- YOU GET
- A process map and a ranked list of what is costing you most — useful even if you stop here.
- FROM YOU
- Access to the team for a few hours, and honesty about the workarounds.
Scope
One win, defined tightly enough to finish.
We pick the single most painful thing and scope it to a first release you could put in front of users in weeks. Everything else goes on a list we keep but do not build yet. This is also where we make the honest build-versus-buy call: if something off the shelf does the job, we will tell you and save you the fee.
- Define the release: in scope, out of scope, why
- Agree the measure of success before we start
- Fixed price, fixed dates, named owner both sides
- DURATION
- 3–5 days
- YOU GET
- A scope document, a price, a date and a success measure. Portable to any developer.
- FROM YOU
- One decision-maker who can say yes to the scope.
Build
Weekly builds you can click, from week one.
Every Friday there is something running that you can open and use. Feedback from that build shapes the next one, which is how a six-week project stays on course instead of arriving as a surprise. Real data goes in early — demo data hides all the interesting problems.
- Weekly deploy to a staging environment you own
- One shared channel, no status-report theatre
- Your data, your infrastructure, from the first deploy
- DURATION
- 4–6 weeks
- YOU GET
- A working system in use, plus the repository and deployment pipeline in your name.
- FROM YOU
- An hour a week to look at the build and say what is wrong.
Prove
Measure it before widening it.
The success measure agreed in scoping gets checked against reality with the people using the thing. Where we have added AI, this is where it earns its place or gets removed: every suggestion stays a suggestion a human accepts, and anything that is not measurably better than the manual version does not ship.
- Compare against the baseline from phase one
- Watch real use and fix what trips people up
- Decide together what phase two is worth
- DURATION
- 2 weeks
- YOU GET
- Before-and-after numbers, and a decision on what happens next based on them.
- FROM YOU
- Willingness to hear that something did not work.
Hand over
You own it, whether or not you keep us.
Code, infrastructure and documentation sit in your accounts from day one, so handover is a conversation rather than a migration. Keep us on a support retainer, bring it in-house, or hand it to another agency — none of those should be expensive decisions, and lock-in is not a business model we are interested in.
- Written handover: architecture, runbook, decisions
- A session with whoever maintains it next
- Optional retainer for support and future phases
- DURATION
- 1 week
- YOU GET
- Everything, in your name: repository, infrastructure, documentation, credentials.
- FROM YOU
- Someone nominated to take the keys.
A typical eight weeks
PHASES OVERLAP — NOTHING WAITS ON A GATE
- Map: week 1
- Scope: week 2
- Build: weeks 3–6
- Prove: weeks 6–7
- Hand over: week 8
Ways to start
ONE DAY
Advisory day
A day of our time on your problem: technical review, build-versus-buy, where AI would and would not help. Ends with written recommendations you can act on without us.
- Best when you are deciding what to do
- No commitment beyond the day
- Covers phase one at a smaller scale
1–2 WEEKS
Discovery sprint
Phases one and two, standalone. We map the process, price the fix and hand you a scope document — yours to build with anyone, including another developer.
- Best when the problem is fuzzy
- Fee credited against a build if you continue
- Ends with a price and a date
6–8 WEEKS
Fixed-scope build
The full five phases on one defined release: mapped, scoped, built in weekly increments, measured against the baseline and handed over in your name.
- Best when the problem is already clear
- Fixed price, fixed dates
- Working software in use at the end
MONTHLY
Ongoing retainer
Support, maintenance and a steady drip of the next phases, at an agreed number of days a month. Cancellable, and never a condition of the original build.
- Best once something is live
- Named days, visible backlog
- Leave whenever it stops paying
How we hold ourselves
01
Small and finished beats big and pending
A tool in use this quarter is worth more than a platform arriving next year.
02
Automation stays a suggestion
Nothing deletes, publishes or approves on its own. A person accepts every AI output, which is why the team trusts it.
03
No lock-in, by design
Your code, your infrastructure, documented. If leaving us is cheap, staying has to be earned.
START WITH PHASE ONE