AI-prepared drafts are starting points. Faster can create task plans, page changes, documents, posts, forms, and customer follow-up, but your team should review the plan and the final artifact before anything is applied.

What this guide helps you do

• Understand the difference between a suggestion, draft, preview, and applied change.
• Edit or regenerate AI-prepared tasks before accepting them.
• Use the owning editor or preview before launch.

Before you start

• Know what the draft will change if approved.
• Have a reviewer who owns the business outcome.
• Ask AI to keep risky work in draft or preview mode when you are unsure.

Do it manually

1. Open the AI-prepared draft, task plan, or preview from chat or the owning workspace area.
2. Read the summary and the exact tasks or changes.
3. Use Edit when a task, assignee, step, or instruction is mostly right but needs adjustment.
4. Use Regenerate when the plan is not the right direction.
5. Use Cancel when the work should not proceed.
6. Accept or apply only after the exact result is clear and review is complete.

Ask Faster AI

• Review this draft and list the changes I should verify before accepting it.
• Regenerate this task plan with fewer steps and no public publishing until approval.
• Compare this draft against the customer context and tell me whether anything looks risky.

Review before saving, sending, or publishing

• Check that task order, dependencies, assignees, and target areas make sense.
• Verify public copy, customer details, dates, pricing, and claims before approval.
• Do not accept auto-generated work that references private implementation details or secrets.

Common issues and fixes

• If the draft is too broad, ask for a smaller plan with one target artifact.
• If the preview does not match the draft, refresh the preview and confirm the latest save.
• If AI chose the wrong target area, cancel and restart with more context.

How Faster keeps AI edits reviewable

For how validators, diffs, checkpoints, and approvals fit together around AI-prepared work, read the AI website guardrails article. Guardrails for AI-Edited Websites: Validators, Checkpoints, and Review.

Versioned workspace model

For the architectural reason draft work can be inspected, revised, and published deliberately, read the versioned workspace article. Why Every Faster Workspace Is a Git Repo.

AI draft approval workflow

For a practical way to decide what gets accepted, revised, regenerated, or rejected, read the AI draft approval article. AI Drafts, Human Approval: Review Workflows You Can Trust.

Faster AI skills overview

For why custom skills should improve draft consistency while review remains the final control point, read the AI skills article. Faster AI Skills: Teach Your Workspace New Tricks.

Meeting write-ups with AI

For why AI meeting summaries still need a human check on decisions, numbers, tone, and follow-up, read the meeting write-up article. Meetings That Write Themselves Up.

AI-drafted proposals

For a four-check review model for AI-drafted proposals before a customer sees them, read the AI-drafted proposals article. AI-Drafted Proposals and Documents (With the Approval Gate).

Meeting-to-follow-up loop

For a follow-up example where the AI draft must be checked for dates, promises, and tone before sending, read the meeting-to-follow-up article. From Meeting to Follow-Up: AI Closes the Loop.

Placeholder media

• Screenshot placeholder: add an annotated screenshot of review an ai-prepared draft with private customer names, emails, amounts, task details, and unpublished content blurred.
• Video placeholder: add a short walkthrough that starts from Workspace, shows the manual path, then shows how to ask Faster AI and review the result before saving, publishing, sending, or applying work.

Related help

• Review AI drafts before publishing
• Review and approve AI page changes
• Create a workflow