Bokili Product4 min read

Every AI Mission Needs a Three-Line Decision Receipt

A finished artefact shows what a learner produced. A three-line decision receipt shows the choice, check and reusable rule behind it.

Bokili EditorialPublished October 4, 2026· Verified October 4, 2026

A completed workplace artefact leading to three connected evidence cards for a choice, a check and a reusable rule

A short AI mission can end with a polished email, summary or plan and still hide the learner’s judgement. The output shows what was produced. It does not show why one option was chosen, what was checked, or what the learner would repeat next time. That gap matters for HR and L&D teams: progress becomes easier to count than skill. A three-line decision receipt fixes the problem without turning a ten-minute mission into an exam.

The idea is simple. After completing the task, the learner records three lines: the decision they made, the check that supported it, and the rule they will carry into the next task. The receipt sits beside the work artefact. A reviewer can then see both the result and the reasoning that made it acceptable.

The output is not the whole performance

Bokili’s guided workplace missions are designed around realistic, bite-sized practice. That format is strongest when the mission names an observable behaviour, not just a deliverable. The US Office of Personnel Management similarly advises training teams to clarify critical behaviours and desired outcomes, then monitor whether they appear in practice. A final file alone may be weak evidence because a fluent AI response can mask a poor choice or a missing check.

Consider a mission that asks an employee to prepare a supplier-risk summary. Two learners may submit equally tidy pages. One copied the first answer. The other separated facts from assumptions, checked a high-impact claim against the source, and flagged the decision that still needs an owner. The artefacts look similar; the receipts reveal very different levels of skill.

The three-line decision receipt

Chose

State the decision made during the task. Name the option, scope or boundary—not the steps taken.

Checked

Name the evidence, rule, risk or acceptance criterion used to test that decision.

Carry forward

Write one reusable rule for the next similar task, in language another colleague could apply.

A receipt is evidence, not a reflection diary

Keep each line specific enough to inspect. “I checked the answer” says almost nothing. “I checked every number against the attached invoice and marked two totals that did not reconcile” creates usable evidence. “I learned to be careful” is vague. “Next time, I will ask who owns the final decision before I draft recommendations” can guide future work.

	Weak receipt	Useful receipt
Chose	I used the best version.	I used the shorter summary because the reader only needs a go/no-go decision.
Checked	I verified it.	I matched all amounts to the source table and marked the unsupported forecast.
Carry forward	Be accurate.	Separate sourced facts, estimates and recommendations before final review.

Reading is a start. Practice makes it stick.

Start learning

The receipt should not reward long explanations. It should make one judgement traceable. That is why three lines work well: they are light enough for frequent practice but structured enough for coaching. They also create a cleaner trail than asking learners to paste full prompts or chats, which may contain irrelevant or sensitive context.

Design the receipt from the behaviour backwards

Start with the decision the mission is meant to teach. If you cannot name it, the mission may still be about completing a feature tour. Then choose one check that would distinguish acceptable judgement from a plausible-looking result. Finally, define a carry-forward rule that can travel beyond the example. This follows the same discipline as teaching one decision rather than one feature, while adding evidence that the decision actually happened.

Build a receipt into one mission

1
Name the target decision
Write one sentence beginning: “The learner must decide whether…”
2
Choose the decisive check
Pick the source, criterion, risk or boundary that a capable learner should examine.
3
Write the three prompts
Ask for Chose, Checked and Carry forward. Keep each answer to one sentence.
4
Test with contrasting examples
Run the mission once with sound judgement and once with a polished but weak shortcut. The receipt should expose the difference.

Use receipts for coaching, not surveillance

A decision receipt is useful only when people know how it will be used. Explain who can see it, how long it is retained and what feedback follows. Evaluate the quality of the decision and check, not a learner’s writing style. If a line reveals uncertainty, route the learner to targeted practice rather than treating uncertainty as failure. The Bokili approach for HR and L&D is built around role-relevant practice and visible progress; receipts can make that visibility more meaningful.

Receipts also connect well with a reusable work artefact and an evidence portfolio. The artefact proves that useful work was produced. The receipt explains the judgement behind it. Over several missions, the pattern shows whether a learner can transfer checks and boundaries across tasks.

Ten-minute mission audit

Choose one current AI mission and underline the decision it is meant to teach.
Draft one sentence each for Chose, Checked and Carry forward.
Imagine a polished but unsafe answer. Check whether its receipt would expose the missing judgement.
Remove any prompt that rewards length instead of evidence.
Add the receipt to the mission and tell learners how reviewers will use it.

The smallest useful record

Training teams do not need a transcript of every interaction. They need enough evidence to see whether practice changed a workplace decision. A three-line receipt is deliberately small: one choice, one check, one reusable rule. When that record sits beside the completed work, a short mission can show more than participation. It can show judgement that is ready to travel.