Skip to content

Mission planning

Mission planning is the routine that turns a pile of signal into today's work. Every morning, for each active mission, it answers two questions: is this mission working, and what should be done about it today?

Why it exists

A mission runs for weeks. Without something looking at it every day, you are the one who has to notice that a milestone slipped, that three prospects raised the same objection, or that nothing has moved since Tuesday, and then work out what to do about it.

Mission planning does that looking so you can spend your time deciding instead of diagnosing. It is the routine to keep if you keep any: its daily diagnosis is the only thing tracking whether your missions are working.

What it does

It works one active mission at a time, and each gets the same two products.

A mission update. A short diagnosis written first, because the plan is built on it:

  • a headline, one sentence of at most eight words, saying whether this week's focus is landing and the mission is on track for its target date;
  • a text of at most thirty words, saying why the signal and the work are or are not converting into movement, and the single bottleneck;
  • a health verdict: on track, at risk, off track or no signal;
  • a one-line status for every pending milestone.

The update reads the trend across recent days, not just today, and it only observes. It does not propose anything. Fresh insights tagged to the mission are its primary new signal. If a mission has had no insight in a week, or is brand new, the routine goes and looks: it reads the connected tool the mission's strategy points at and runs a couple of web searches, and saves what it finds as insights.

It may also fold what changed into the mission's playbook, but only when direction or a constraint durably changed. A one-day blip does not rewrite your strategy, and most days it writes nothing there. It marks a milestone completed only when what it read proves it landed.

A day plan. One day plan per mission: the handful of tasks worth doing, each with the reason it is there. The mission's intensity decides how wide and bold the plan is.

Every proposed task has to pass three checks, or it is dropped:

  1. It produces an output the agent can author itself: research, drafting, analysis. Publishing, sending and anything that needs a human (calls, meetings, configuring another tool) are yours, and a task that mixes both is split so the proposal is only the half the agent can do.
  2. It is not already shipped, in flight or scheduled.
  3. It advances this mission, and its rationale says how.

Each rationale carries its receipts: the insights it rests on and the milestone it should move, so you can check the reasoning before you approve. A task can be dated for a later day when both the deliverable and the date are certain, such as Thursday's draft for Friday's changelog.

The plan can also carry up to three open questions, only when the answer would change which tasks belong in it. Most days there are none.

What you see

One card in your inbox per mission that has something to decide. Approve the tasks worth doing and reject the rest. Approving turns them into real tasks and starts a follow-up run that drafts each task's prompt and schedule. The plan is settled from that point.

Things worth knowing:

  • No card is not a failure. A mission with no tasks that pass the checks and no questions is recorded for the day and posts nothing.
  • A new card replaces yesterday's. If an earlier plan is still open when today's arrives, it is closed, and its unactioned tasks are carried into today's card only where they still matter, with today's rationale.
  • A failed task comes back as a decision. If it failed, the routine either re-proposes it with a note on what changed or drops it when the capability is missing.
  • Scheduled work stays true. A scheduled task that this week's focus or today's signal now contradicts is moved or cancelled, with a reason on the task. Both show in its History, where you can undo them.
  • A passed target date asks you. If an active mission is past its target date, the routine does not move the date. It plans the day as usual and asks on the card whether to extend, close it as done or drop it.

When it runs

Weekdays, in the morning. It runs after the extraction routines, so it sees the signal they just captured.

It plans only active missions, so this is where mission status earns its keep: a workspace with nine active missions gets nine plans every morning. If there is no active mission, the run is skipped.

Tuning it

You cannot edit the routine, but you can steer it:

  • Status decides which missions get planned at all. Pause the ones you are not working this week.
  • Intensity sets how hard each mission is pushed.
  • The playbook is what the planner reads first: strategy, cadence, rules and the roadmap. A sharper playbook gives a sharper plan.
  • Your answers to the insight request become the facts the next diagnosis uses.