Skip to content

Level 3 · Guide 14

Design human approval into Claude workflows

A person must review context and the proposed result before automation spends money, sends messages, publishes, changes permissions, or makes an irreversible write. Automate reading, classification, and drafting, but stop at preview and approval pending.

Product features may change. Check the last verified date and official sources.

Core concepts to know first

Focus on the decisions these terms support in real work rather than memorizing them.

Core concepts

1. Approval is a system state

Approval is a separate state, not a phrase inside a prompt. The system should present the exact target, scope, expected effect, evidence, and recovery option, then wait without performing the action. In this guide, the first observable move is to list every action that creates an external or hard-to-reverse effect.

2. Classify actions by consequence

Define the preview evidence and responsible approver for each action. Preserve the approved input and the evidence behind the result so a reviewer can distinguish what the source says from what Claude inferred. This directly controls the risk that a vague approve button can hide the target or scope.

3. Preview target, scope, and effect

Test that the workflow stops at draft, preview, or approval pending. Record the decision and the remaining uncertainty instead of hiding it in polished prose. The review must explicitly test whether approval can become stale after inputs change.

4. Separate maker and approver where needed

Treat the possibility that the same person creating and approving can weaken control as a required test case. The intended result is an approval-gate matrix covering action, impact, owner, and recovery, not an unreviewed answer that merely looks complete.

5. Design rejection, expiry, and recovery

For Design human approval into Claude workflows, the final concept joins the earlier checks into an operating boundary: use approved inputs, expose evidence and uncertainty, and stop before a consequential action. Ownership and a reproducible review determine whether an approval-gate matrix covering action, impact, owner, and recovery may move beyond training.

Synthetic work scenario

How this applies at work

This scenario was written for learning and is not a real customer case.

A fictional procurement workflow drafts a vendor comparison and purchase request from synthetic data. It displays amount, target, and assumptions, then stops; a responsible person completes any real action through a separate process.

The scenario is newly written for this guide and is neither a customer case nor a performance claim. Its specific deliverable is an approval-gate matrix covering action, impact, owner, and recovery. A reviewer can reproduce the work from the synthetic inputs without access to customer, employee, health, contract, or confidential company data.

Try it yourself

Choose one small task and follow the steps. Confirm organizational policy and data boundaries before using sensitive materials.

  1. Step 1. List every action that creates an external or hard-to-reverse effect

    List every action that creates an external or hard-to-reverse effect. Use only approved synthetic material and record both the evidence and any remaining uncertainty.

    Verify: Confirm that another reviewer can reproduce the input, result, evidence, and stop point.

  2. Step 2. Define the preview evidence and responsible approver for each action

    Define the preview evidence and responsible approver for each action. Use only approved synthetic material and record both the evidence and any remaining uncertainty.

    Verify: Confirm that another reviewer can reproduce the input, result, evidence, and stop point.

  3. Step 3. Test that the workflow stops at draft, preview, or approval pending

    Test that the workflow stops at draft, preview, or approval pending. Use only approved synthetic material and record both the evidence and any remaining uncertainty.

    Verify: Confirm that another reviewer can reproduce the input, result, evidence, and stop point.

Completion checklist

Check only what you verified yourself. Every item must be checked before saving completion.

Completion checklist

Some items are still unchecked. Review the result again.

What could go wrong?

Plausible language does not guarantee accuracy. Compare the result with originals, calculations, permissions, and current information.

  • A vague approve button can hide the target or scope
  • Approval can become stale after inputs change
  • The same person creating and approving can weaken control

Boundaries that require human review

Academy practice stops at draft, preview, or approval pending. Actions with real impact require separate owner approval outside Academy.

What AI can do

  • Step 1. List every action that creates an external or hard-to-reverse effect
  • Step 2. Define the preview evidence and responsible approver for each action
  • Step 3. Test that the workflow stops at draft, preview, or approval pending

What a person must approve

  • Applicable law, contract, organizational security policy, and explicit approval boundaries take priority. Stop and ask the responsible owner when they conflict.
  • Academy exercises never send, publish, purchase, delete, execute contracts, or change permissions. A responsible person performs any real action through a separate process.

This guide is educational and does not replace legal, security, or privacy judgment for your organization.

Questions about this guide

When is this guide complete?
It is complete when the stated outcome is ready and another reviewer can retrace the inputs, evidence, boundaries, and decision. The target outcome is “An approval-gate matrix covering action, impact, owner, and recovery.”
May I practice with real company data?
No. Use synthetic material in the Academy. Real data requires a separate review of policy, legal basis, contracts, minimization, retention, deletion, and the approved environment.
What if the product screen differs from this guide?
Product behavior can change. Check the verification date and official sources, then retest the current account and plan with a small, low-risk example.

Official sources and further reading

Recheck the current product and policy status in these primary sources.

  1. Anthropic AI Fluency Index · Anthropic Research
  2. Get started with Claude Cowork · Claude Help Center
  3. Security · Claude Code Docs

Authorship and review

Author
QJC
Last verified
2026-07-23
Update sensitivity
High
Tested product surface
Official documentation and QJC training scenarios using synthetic data
Tested plan
Reconfirm feature and account availability in official sources on the day of use

Save progress

Completion and checklist items are saved only in this browser. They do not sync to other devices or browsers.

Academy does not collect or store work materials, prompts, or outputs. It runs no separate analytics scripts beyond basic server access logs.

QJC Claude Academy is unofficial educational content independently created and operated by QJC. It is not an official course operated, sponsored, certified, or affiliated with Anthropic. Claude and Anthropic are trademarks of Anthropic PBC.