For our first two years Keel had a planning meeting every other Monday. It took 90 minutes, involved everyone, and mostly consisted of copying unfinished tickets from one sprint to the next. In January 2025 we cancelled it. Nobody has asked for it back.
This is not an argument against planning. It is an argument against doing by hand the parts of planning a tool can do, so the people can do the part only they can: deciding what matters.
What the cycle does on its own
- Rolls over unfinished work. At midnight on the last day, open issues move to the next cycle with their history.
- Fills to capacity. It pulls from the top of the prioritized backlog until it reaches the team's average of the last three cycles.
- Flags risk early. Anything blocked for two days, or any project whose scope grew more than 20%, shows up on the cycle page.
Cycle 24 Nov 2 – Nov 13
scope done
What people still do
The backlog order is the plan. Keeping it honest is the one job we refuse to automate. Each project lead spends about twenty minutes on Friday reordering their part of it, and that is the whole planning ritual.
“We cancelled our Monday planning meeting in March and nobody has asked for it back.”
Brightloom went further than we did. Their median cycle time has gone from 5.1 days in January to 2.5 in September, mostly because issues stopped waiting for a meeting to be picked up.
When a meeting still helps
Big bets. When a project will take three cycles or more, we still sit down for an hour to cut it into milestones. The cycle can fill itself; it cannot decide that offline mode matters more than search this quarter.