AI review

Use AI output as a reviewable suggestion while keeping people accountable for customer-facing action.

Expected outcome: The final summary, draft, or recommendation is supported by authorized context and approved through normal product controls.

1. Make AI useful without confusing suggestion with fact
WRKZY can help summarize, draft, and recommend. A person remains responsible for evidence, judgement, and action.

2. Confirm the context is appropriate for the task
Use only customer and workspace information the current role is allowed to access. Avoid unnecessary sensitive data.
Review point: Access to a feature does not expand data permission

3. Read the underlying record before the suggestion
Review the full thread, customer identity, timeline, consent, deal, quote, and recent commitments that affect the answer.
Review point: The source of truth remains the product record

4. Generate a narrow summary, draft, or recommendation
Ask for the specific job and audience. Treat the result as uncommitted working material.
Review point: Nothing should send or update merely because text was generated

5. Check facts and customer commitments
Compare names, dates, amounts, products, delivery claims, policies, next actions, and quoted language with the underlying evidence.
Review point: Plausible wording can still be unsupported

6. Check tone, fairness, privacy, and business judgement
Ask whether a reasonable teammate would make the same recommendation with the same evidence.
Review point: Edit or reject—do not rationalize a weak suggestion

7. Use the normal control to perform the action
A human reviews the final recipient, content, fields, or workflow and explicitly sends, saves, or applies it.
Review point: Normal permissions and audit history still apply

8. Keep the final action and its evidence connected
Record corrections and feedback when available. Never use remembered AI output as a substitute for current customer context.

This silent visual walkthrough was screen-recorded from the sanitized Acme Retail demo workspace. Follow the linked written guide for exact prerequisites, role boundaries, edge cases, and recovery.