9order LEARN 9order ↗
Day 1 / Session 1 / 120 minutes

Authorise

What useful work may AI actually do?

Session / guided simulation Guided simulation · AI/GPT not required
What you'll learn

Learn to separate conversation output from real-world action, assign approval to a specific person and preserve useful preparation when authority is missing.

Bring / required

Required: your AI Work Delegation Blueprint and either your own workflow or the fictional starter case. AI/GPT is not required for the core simulation.

What you'll do

Use the starter records or your own workflow to sort drafting, approval, sending and follow-up into distinct actions, then record the allowed boundary.

Take away

An action boundary, named approver and completion evidence added to your Delegation Blueprint.

This session uses paper, partner role-play and supplied evidence. Do not treat the simulation as a live customer action.

What you will leave with

An action boundary and a named approver. Add this to your AI Work Delegation Blueprint.

Use your own workflow or the fictional quotation case. No coding or live customer action is needed.

Start with a decision · 10 minutes

An assistant says “Your quote is ready.” Has it drafted a message, sent it, or committed the business to a delivery date? Ask everyone to choose before showing the records.

Bite A · Separate an answer from an action

Question: What changed outside the conversation?

Input: S0, S2 and the first three S6 state cards. Open the case cards.

Try · 12 minutes: Sort drafting Q-104, sending it, and accepting its delivery date. Circle the first external change. Describe one useful action in your own work or use the supplied quote.

Compare after you have decided

A draft helps the coordinator. Sending changes what the customer receives. Neither step confirms stock or a delivery date.

Keep: Target, allowed change and completion evidence.

Bite B · Give authority to a specific action

Question: Who approves this version for this recipient?

Input: S6: v1 unapproved; Mira approves v2 at 10:03; M-88 records its send at 10:05. Open the case cards.

Try · 12 minutes: Role-play assistant, Mira and observer. Present the exact draft and recipient for approval. Move the paper record only after approval; observer records what changed.

Compare after you have decided

The receipt supports one approved send. Approval of v2 is not standing permission for future messages. This exercise is simulated.

Keep: Action mode, approver and approval scope.

Bite C · Keep useful work moving at the boundary

Question: Can the quote progress while the discount waits?

Input: Q-106 asks for 15% off; S2 gives Mira up to 10% and Daniel above 10%. Open the case cards.

Try · 12 minutes: Prepare the undiscounted quote and an escalation to Daniel. Identify what pauses and what can continue. Compare the consequence and reversibility of drafting versus sending.

Compare after you have decided

The assistant may prepare. It may not approve the discount; Mira cannot authorise 15% under this policy either.

Keep: Permitted preparation and human decision.

Put it together · 20 minutes

A partner can identify the exact action, target, mode and human authority. Record a useful preparation step even when sending remains blocked. Revise the weakest item and have a partner use it. Keep the result and one limitation.

Close your question · 10 minutes

Return to What useful work may AI actually do? and your own opening question. Answered, partly answered or still open? Show the evidence. Save the updated AI Work Delegation Blueprint. A reasoned decision to keep a step human-led is useful work.

Common trap

Do not equate technical ability with permission, or label a role-play as a live action.

Do I need connected tools? No. Paper state cards demonstrate the decision; a later sample can show the actual controls.

Where this appears in real work

Lead-to-appointment: suggesting a slot, recording consent and making a booking are different actions. The class uses the fictional packet; selected project demonstrations and technical walkthroughs can add detail later.

← PreviousCourse mapNext →