Skip to main content

Authorise one bounded job

On this page

Authorise one bounded job

Put this to work

Draft the bounded-work mandate

Before lead qualification

Copy this prompt. Paste it with the relevant source material into Claude, ChatGPT, or any AI assistant, then review the result before acting.

Help me prepare a bounded-work mandate for one lead-qualification trial. Do not expand the authority I provide.

Job: Qualify one received sales lead using only supplied evidence.
Beneficiary: [possible buyer and accountable seller]
Accountable human owner: [name or role]
Worker or assistant: [person or AI tool]
Allowed information: [exact supplied sources]
Allowed actions: Read the supplied evidence and draft a Lead Qualification Brief.
Prohibited actions: External lookup, enrichment, buyer contact, CRM mutation, commercial promise, or final qualification decision.
Time/resource limit: [for example, one draft within ten minutes]
Approval point: The owner reviews the brief before any next action.
Stop conditions: Invented facts, hidden unknowns, unclear ownership, permission expansion, or requested external action.
Recovery: Preserve the draft and return the lead to manual assessment.
Review time: [date and time]

Return a concise mandate with: outcome, owner, worker, permitted inputs and actions, prohibited actions, limit, approval, stop, recovery, evidence destination, and expiry. List any ambiguity instead of filling it in.

Next →Choose the right model and tool for this prompt

UNTESTED · claim_state: UNTESTED · outcome_maturity: UNTESTED

Candidate teaching projection; contextual recipe unproved. The JTBD Ledger keeps this route as a gap until a contextual owner accepts the method and retains decision evidence.

Workflow family: People, Capability & Governance.

GATE

  • Authority: human_authority: Accountable mandate owner
  • Beneficiary: people affected by the work and the accountable human owner
  • Permitted inputs and actions: only the recorded information, tools, spend, contacts, and actions in the accepted mandate
  • Prohibited: implied permission, scope expansion, unapproved contact or mutation, and consequential action without the named approval
  • Stop: stop_rule: Stop for ambiguous or expired authority, protected-constraint risk, external consequence, or failed recovery.

CHECK

For every mandate condition, record exactly one evidence state: supported, contradicted, or unknown. Never infer permission from access, silence, title, or a completed task. Unknown authority means the work does not start.

NEXT MOVE

Have the accountable human accept or correct the mandate. Only then run the featured Qualify a Lead job, and return its evidence to the named review point.

Outcome​

The worker and people affected receive an accepted mandate stating the outcome, responsibility, allowed actions and information, approvals, evidence, stop, recovery, and correction owner.

Use When​

Trigger this job before a person or agent begins delegated work. The accountable human mandate owner retains purpose, permission, spending, accepted risk, consequential judgment, authority changes, and stop.

Inputs​

Bring the job and beneficiary, desired output and outcome, named worker, allowed tools, data and actions, prohibited actions, resource and time limits, approval points, gauge, stop conditions, recovery path, and review time.

Recipe​

  1. Name one job, beneficiary, owner, and observable acceptance condition.
  2. List permitted and prohibited actions, information, tools, spend, and contacts.
  3. Define checkpoints, escalation, stop conditions, and a recoverable state.
  4. Ask the worker to restate the mandate and expose ambiguities.
  5. Record human acceptance before work starts; review authority when conditions change.

Human and AI Control​

An agent may act only inside the recorded mandate. Technical access is not authority, silence is not approval, and completion does not broaden permission. Stop on ambiguity, protected-constraint risk, expired authority, unexpected external consequence, or inability to recover.

Output and Handoff​

Return the accepted mandate, acknowledgements, approval state, checkpoint, evidence destination, stop owner, and recovery instructions. Hand authorised work to its job owner and the resulting receipt to outcome review.

Evidence and Stop​

Retain the mandate revision, restatement, approvals, actions, exceptions, stops, and recovery outcome. Compare actual action with permission and acceptance. Stop and contain work when scope or authority diverges. A signed mandate does not prove correct execution, safety, or beneficiary value.

Next​

Correct the mandate when a boundary was misunderstood or insufficient. The human owner alone expands authority; return execution-method failures to the contextual job and governance failures here.