From advice to action: turn counsel into commitments and learning

Name the decision before the task

A task without a decision can become motion that preserves ambiguity. Write what was chosen and which trade-off was accepted. If the council concluded that more evidence is required, name the evidence decision: what will be learned, by when, and how that result will change the larger choice.

Keep counsel and ownership separate. The mentor may have proposed an action, but the user must choose it. Record whether the commitment was accepted, modified, or rejected. This avoids later presenting generated advice as an instruction the user had no role in authorising.

Write a reviewable commitment

Use a concrete verb, visible output, owner, deadline, and success signal. “Work on positioning” is not reviewable. “Interview five buyers in the selected segment by Friday and record the words they use for the urgent problem” creates evidence. Define what completion means without pretending the external outcome is controlled.

Attach the commitment to the mission and decision it serves. That connection matters when priorities change. A task can be completed perfectly after its reason has disappeared. The system should allow the user to revise or cancel deliberately rather than reward completion detached from current purpose.

Choose the right follow-up

Assign follow-up based on the failure mode the seat is meant to notice. A strategy mentor may ask whether the evidence changed the assumption. An execution mentor may ask whether ownership or process failed. A psychological lens may examine avoidance or attention. The Meta-Advisor coordinates timing so several mentors do not send duplicate pressure.

Let the user control channel, frequency, quiet hours, and communication mode. A useful follow-up can be a single direct question with quick replies: done, blocked, changed, or no longer relevant. It should not use manufactured intimacy, escalating guilt, or threats to engagement metrics.

Diagnose a miss before rescheduling

When a commitment is missed, identify the class of block: unclear next action, inadequate time, missing resource, dependency, fear or avoidance, changed evidence, conflicting goal, or a commitment that was never genuinely accepted. Each class suggests a different response. Repeating the same deadline does not solve a structural block.

Revise the plan at the smallest useful level. Clarify the first action, reduce scope, obtain a resource, renegotiate a dependency, schedule a conversation, or close the commitment. Preserve the reason for revision. Over time, those reasons can reveal recurring gaps in planning, environment, or council composition.

Review outcome and process

At completion, compare the success signal with what happened. Capture the result, unexpected information, stakeholder response, and next implication for the mission. Do not force a success narrative. A well-run experiment that disproves the leading assumption can be more valuable than a completed task that taught nothing.

Ask which advice was used, which was ignored, and whether the citations genuinely supported the application. Update the mission brief and the status of the original decision. This closes the loop from source to interpretation to action to evidence—the sequence that makes a mentor system progressively more useful.

Keep a compact commitment ledger rather than relying on conversation history. For each item, preserve the originating decision, owner, due date, latest status, reason for any change, and the evidence produced. Review the ledger at the mission level: repeated cancellations may indicate weak priorities, while repeated blocks around the same dependency may justify a new council seat or an operating change. The purpose is not to maximise task completion. It is to distinguish useful adaptation from avoidable drift and make the next commitment more realistic.