AI Estimation Reveal Rounds for a Fictional Craft Project

As of 2026-10-23 · The AZET team. Written by the AZET team as a creative reference exercise for fun and practical exploration. All examples are hypothetical, not evidence of AZET product capabilities. This is the article reference date, not a current product verification date.

Three folded slips show 2, 5, and 8. They refer to the same fictional task: make a hanging paper moon for a miniature night garden. One person imagined a flat circle, another a folded ornament, and the third a hollow sphere.

In this reference exercise, the numbers represent relative making effort. They are not hours, prices, or commitments. The goal is to expose different understandings of a craft task while preserving each person's independent first estimate.

Define the unit before hiding the slips

Choose one small reference task: cut a supplied star outline from plain paper, punch one marked hole, and attach a supplied thread. Call that size 1.

Offer the values 1, 2, 3, 5, and 8, plus “unclear.” Define effort broadly enough to include unfamiliar assembly and fiddly steps, while keeping material cost outside the exercise. Do not tell participants that an 8 means exactly eight times the duration of a 1.

Give everyone the same task description and finish condition. For the moon, the starting description deliberately says only “make a hanging paper moon that can be viewed from either side.” Participants write a size and one sentence about what they believe they are making.

Make the first round genuinely independent

Ask each person to record an estimate privately before discussion. Reveal all slips together. If someone announces a number early, pause and record that the round no longer offers an independent starting point. Run a fresh practice task rather than pretending that resealing the slips removes what people heard.

In a proposed AI-assisted version, keep the drafting request separate from the first estimate. The prompt can prepare a neutral task packet or a comparison sheet for later review. Do not show a suggested size before participants have written their own; that would insert another visible estimate into the exercise.

No one needs to defend their competence. The explanation concerns the imagined object and work steps. “I counted shaping six curved panels” is useful. “I am slow at crafts” does not tell the group whether it understood the task consistently.

Work the moon through two reveals

In the authored first round, Mira records 2 for a flat double-sided circle. Tomas records 5 for two curved panels joined at the edges. Jules records 8 for a sphere assembled from multiple pieces. Keep these explanations beside their original numbers.

The group clarifies the task: a two-panel folded moon, made from supplied outlines, with one hanging thread. It should hang freely and present a finished face on both sides. This wording excludes the flat circle and the sphere without selecting a size.

Each participant now estimates privately again. Imagine the second reveal gives 3, 5, and 5. Mira assumes the supplied fold marks align. Tomas and Jules expect an adjustment pass to make the panels meet neatly.

The group records “3 to 5, alignment untested” and chooses a paper mockup as the next learning step. It does not average the estimates into a supposedly precise result. Nor does the second round erase the first: the change from three different objects to one shared object is the main evidence produced by this fictional exercise.

An exact prompt for reviewing the revealed slips

Use the prompt after the independent round, with only the supplied fictional material:

Review a fictional craft estimation round. Sizes are relative effort,
not hours or money. Reference size 1: cut a supplied star outline,
punch its marked hole, and attach supplied thread.
Task: make a hanging paper moon visible from either side.
Revealed estimates:
Mira: 2, flat double-sided circle.
Tomas: 5, two curved panels joined at the edges.
Jules: 8, sphere assembled from multiple pieces.
Preserve each number and explanation. Identify differences in the
imagined object, supplied materials, and assembly assumptions.
Draft one neutral clarification for each difference.
Do not select a winning estimate, calculate an average, or infer skill.
Return a short task-clarification sheet for human review before round two.

Inspect any proposed clarification for a hidden answer. “Confirm that the easy flat version is intended” favors both an object and a size. Replace it with wording that lists the unresolved object forms without ranking them.

Check the stopping condition

Set a limit of two reveal rounds for this practice session. If the descriptions still differ, leave the estimate unresolved and identify the missing detail. Repeating rounds until everybody chooses the same number can conceal uncertainty rather than examine it.

Before closing, check that every estimate still points to its own round and task wording. Confirm that “unclear” remains an allowed response. Record the small experiment that would inform a later estimate, without changing that proposed experiment into an assigned commitment.

FAQ

Must everyone agree on a single size?

No. Keep a range when people still expect different assembly work. Attach the unresolved assumption so a later mockup has something specific to test.

Should the highest and lowest estimators speak first?

They can explain their assumptions first, provided the group examines the task rather than judging the people. Let anyone with a distinct assumption contribute even if their number matches another slip.

Can we compare sizes across different craft groups?

Only with care about their reference tasks and definitions. This exercise gives one group a local comparison unit; it does not establish a transferable measurement scale.

E1, our maker introduction dated October 3, 2026, described an assistant and an early-access waitlist. The reveal-round exercise is original reference material and does not establish an AZET estimation capability.

To register your interest through the AZET waitlist, visit azet.io.