Start with the real problem. Make useful progress early.
The process is intentionally lightweight: understand the outcome, get close to the actual evidence and systems, act where useful, show what changed, and keep the next decision clear.
Bring reality
Send the outcome, question, system, repository, evidence, prior attempts, and any deadline or constraint that changes the answer.
Find leverage
Identify the highest-value technical boundary instead of assuming the visible symptom or first requested implementation is the real problem.
Work in the system
Research, debug, design, write code, automate, evaluate, or produce the technical artifact that actually moves the outcome forward.
Show evidence
Make changes and conclusions inspectable through tests, readback, changed artifacts, screenshots, measurements, or decision-ready reasoning.
Continue intelligently
Use what the first work teaches us. Stop when the result is enough; expand only when the next work is genuinely worth doing.
What helps us move fast
- Access to the actual source/evidence when appropriate
- A clear description of why the outcome matters
- Known constraints and unacceptable failure modes
- Permission to question the proposed solution when the evidence points elsewhere
What you should get
- Useful movement on the real problem
- Technical reasoning you can inspect
- Changed code/artifacts where implementation is part of the answer
- A clear current state and next decision instead of process for its own sake