Work
New Discovery Process
Why I built it
Discovery kept collapsing into stakeholder theater — decks about assumptions instead of evidence. I rebuilt the process as a working system: capture, test, commit. These are the three mechanics that changed behavior.
Year
'26
Client
NDA
Service
Systems Orchestration
The build, live
Don’t read about it. Click it.
Ambient sync
Off
Manual only — nothing moves without you.
01
Assumptions go live
An assumption isn’t real until it’s switched on and tracked. This toggle made ‘we think’ and ‘we’re testing’ two different states.
02
Evidence, three ways
The same research re-cut by signal, adoption, and cost. Arguments got shorter because the lens was shared.
Signal
Adoption
Cost
Weekly signal volume, normalized.
Capture the intent
One field, no form. The system infers the rest from context instead of asking.
Back
Next
03
Commit is a walkthrough
Before anything ships, the flow replays the decision back in plain language. Nobody signs off on something they didn’t understand.
Decision log
What I chose, and what it cost.
Kill the discovery deck
Decks reward narrative polish over evidence. A live system rewards being right.
Executives missed the theater at first — the system had to earn the room.
Timebox every assumption
An untested assumption older than two weeks is a belief. Deadlines force resolution.
Some genuinely slow questions got forced into premature answers.
Decisions are written before they’re made
Writing the commit doc first exposes the gaps while they’re still cheap.
Feels bureaucratic for small calls — we exempted anything reversible in a day.
Artifacts