Harvesting inventions that already exist

A cycle that works

Quarterly sessions per team, one hour, small groups. Five questions: what did we try first that failed and why; what turned out to be much harder than expected; where did we depart from the standard approach and what forced it; what would a competent competitor get wrong building this from our datasheet; what did we measure before and after where the after was surprisingly better.

Capture the answers in the format a drafter needs - problem, mechanism, parameters, alternatives, boundary conditions, data - not as a list of achievements.

Triage, and timing

Triage commercially. For each candidate: what would a competitor have to do differently if it were granted, and could we detect them using it? Candidates that fail both belong in trade secrecy or nowhere.

Timing is the part organisations get wrong. A harvesting session scheduled after a product launch is archaeology - the disclosure has already happened. Tie the cycle to the development calendar, before public milestones.

What to actually do

• Ask about problems solved, not features delivered.
• Run sessions before public milestones, not after.
• Capture in the drafter format, not as achievements.
• Triage on competitive impact and detectability.