What's true?
- Current-state map
- Data-gap analysis
Ground truth
Stakeholder interviews, a system inventory, and a data-lineage walk — against what is running, not what the deck says.
No assumption survives
Principle Pattern
Every project starts by finding out what's actually true.
Most don't. They start with a plan built on what the last deck claimed, and every decision after that inherits it. Discovery is the four weeks where we go and check — your systems, your data, your risk, your numbers — and hand you the evidence either way.
- Four weeks
- Eight artifacts
- Yours to keep
The expensive part isn't building the wrong thing — it's the nine months before anyone admits it.
Projects don't fail at the end. They fail in week one, quietly, when a plan gets built on a claim nobody checked — and then every decision after that inherits it. Discovery is the deliberate act of checking, in the order of how expensive it would be to be wrong. Four weeks of it buys back the nine months.
“We tell clients what they need to hear before what they want to hear.”
The Sprint Track
Four weeks, and what each one hands you.
What breaks?
- AI-risk register
- Responsible-AI map
Map the risk
A failure-mode workshop and a responsible-AI review. Where this breaks, who it hurts, what the auditor will ask.
Named, not hinted at
What's worth it?
- Ranked use-case backlog
- Target-state architecture
Model the value
Use cases scored on value, feasibility and risk. Target-state architecture drafted against your real constraints.
Scored, not guessed
What now?
- Defendable ROI model
- 90-day plan
The case for change
ROI with its assumptions stated, a build-buy-skip call, a 90-day plan, and the honest list of what not to do.
Signed off, or stopped
Optional
Weeks 5–6, if the decision needs proof
- Proof-of-concept build
- Executive readout
The Kit
0 of 8 deliverables.
- Current-state map
- Data-gap analysis
- AI-risk register
- Responsible-AI map
- Ranked use-case backlog
- Target-state architecture
- Defendable ROI model
- 90-day plan
The Kit
What you walk away with.
Eight artifacts, in source format. Not a slide deck summarising them — the artifacts. Open any one to see what is actually in it.
Risk Reversal
What you keep even if you never hire us.
Every artifact above is yours, in source format, licensed to you outright. No watermark, no “sample,” no dependency on us to read it. Take the architecture to another firm. Take the ROI model to your board. If the honest answer is that you should build this in-house, that is what week four will say.
We have recommended “don't build this” on 0 of our last 0 sprints.
Who's in the room
The people in the kickoff are the people in the readout.
- Principal architect
Writes the target-state architecture and defends it to your engineers at the readout. If it does not survive that conversation, it was not finished.
Named and introduced before you sign.
- Responsible AI lead
Classifies the system against the risk tiers that apply to it, writes down the reasoning rather than just the tier, and names the failure that would have reached a headline.
Named and introduced before you sign.
- Value and economics
Builds the business case to be attacked by your CFO, with the assumptions listed separately from the arithmetic so both can be argued with.
Named and introduced before you sign.
What it costs
Fixed price, agreed before week one.
From a scope we write together. Two things move the number, and neither of them is a tier you have to pick from a table.
What's fixed
- The price, once agreed
- The four-week timeline
- All eight artifacts
- The three people on it
What moves it
- How many business domains are in range
- Whether you want the proof-of-concept in weeks five and six
Scope changes are a new conversation, not an invoice you did not expect. The number itself comes out of a thirty-minute scoping call — it is the one thing on this page we cannot write down in advance, because it depends on answers we do not have yet.
Book a scoping callPattern Prototype
Three of these eight artifacts are a design brief in disguise.
Discovery tells you what to build and in what order. It does not make the thing legible to the people who have to trust it — that is a separate discipline, and it is the one that decides whether any of this gets adopted.
Ranked use-case backlog
Decides what gets designed first — and what ships as a rules engine with no interface at all.
Target-state architecture
The constraints the interface has to honour, so the prototype survives integration.
Responsible-AI map
Becomes real surfaces: explainability, confidence, the moment a human takes over.
Discovery stands on its own. This is what we'd do next, if you want it.
See the Design phaseBefore we meet
Come to the call already halfway through it.
A one-page Discovery Brief — twelve questions we'd ask anyway. Fill it in beforehand and the first session starts at depth instead of introductions.
Download the Discovery Brief(PDF, 1 page)Ready to find out what's actually true?
Four weeks, fixed price, senior architects only. You leave with eight artifacts in source format — whether or not we build what comes next.
The Usool Method
- Principle
- Pattern
- Prototype
- Production
- Proof