From making to directing and judging
In The Kite Method no roles disappear, but the centre of gravity shifts from making to directing and judging. Each role gets a Kite name, so the organisation can see the work itself has changed.
Pilot
Writing Frames, steering agents, judging outcomes; coding hands-on in the Paired and Held zones
- Holds the line on
- Technical quality of their own work
Intent Owner
Quality of Kite Briefs and acceptance criteria; closer to the build than before
- Holds the line on
- The why and the priority
Tail Engineer
Designing verification: test harness, evals, quality gates in CI
- Holds the line on
- Whether “done” really means done
Frame Architect
Capturing architecture rules as boundaries agents can read; decides in the Held zone
- Holds the line on
- Irreversible technical choices
Line Keeper
Guards zones, review load, Human Anchors and people's growth
- Holds the line on
- Health of the team and the method
Combining roles in small teams
In small teams one person often combines several roles. That is fine, with one rule: whoever built an item, or had it built, is never the only one who checks it in the Guided and Held zones.
Team size
The Kite Method works from two people upwards, but how you apply it depends on team size. The boundaries below are guidelines, not hard rules.
Small team
Everyone is a Pilot; Intent Owner and Line Keeper are combined. Nobody is the only one checking their own Guided or Held work.
The Kite Retro merges with the Line Review.
Standard team
Every role filled separately. This is the size the method is designed for.
All three rhythms run separately: Daily Demo, Line Review, Kite Retro.
Multiple teams
Each team has its own Pilots, Intent Owner and Line Keeper. One Frame Architect guards the architecture rules across teams.
Each team runs its own loop and zones. Shared: one Tail (test infrastructure and evals) and a joint quarterly Kite Retro to keep zone routing consistent.
What AI does to team size
Teams become smaller but broader. Teams used to grow to gain build capacity; now review capacity is the limit. A team is big enough when it can check all Guided and Held work with real attention. If review load rises structurally, that is the signal to strengthen the team or improve the Tail, not to approve faster.