Guide
How to run an AI PRD review with PRD Scan
Run an AI PRD review in Draftlytic. Get a quality score and up to 10 concrete fixes for gaps, vague requirements, and missing details — then apply them in one click.
What it is
Before you hand your spec to Cursor, Lovable, Bolt, or whatever you're building with, it helps to know whether it will actually produce good code — or whether the AI coding tool will start filling in gaps on its own. PRD Scan runs an AI PRD review on your spec and gives you a quality score (Weak, Good, or Excellent) plus a prioritized list of up to 10 concrete fixes. It tells you exactly what is missing, vague, or inconsistent, so you can fix it before the code starts.
The scan is available from two places: the project detail page and the PRD Workshop. Both use the same backend, so the quality of the analysis is identical. On a saved project it reads your full structured spec — features, data model, navigation, business section, and everything else you've filled in. In the PRD Workshop it reads the free-text PRD you are working on. Either way, the AI tailors its feedback to your chosen AI coding tool, so the fixes it suggests are the ones that actually matter for the tool you're using.
Scan costs 30 credits per run and is available on all paid plans (Starter, Plus, Pro).
Step by step
Open your project
Navigate to any saved project on your dashboard. The project must be in a ready state — a project still generating won't have the scan option available.
Click Scan with AI
In the project header toolbar, click the Scan with AI button. The button shows a credit cost indicator so you can see the 30-credit charge before you commit. If you are on the free tier, clicking the button takes you to the pricing page instead.
Confirm the scan
A dialog opens showing your current target tool — the AI coding tool you have set in your profile (Cursor, Lovable, Bolt, Windsurf, or a custom one). If no target tool is set, the scan runs in tool-agnostic mode. Click Run scan to start. Once the scan is running, credits are deducted and leaving the dialog does not refund them.
Review your score and fixes
The scan returns a score (Weak, Good, or Excellent) with a short rationale, then a numbered list of up to 10 suggested fixes. Each fix has a title and a description explaining exactly what needs to change and why. If you are on the Starter plan and the scan flagged acceptance criteria or navigation endpoint gaps on a project created at Brief or Standard depth, you may also see a callout suggesting Detailed depth for your next project. Plus and Pro users already have Detailed depth available, so the callout is not shown to them.
Open the results in the project editor
Click Open in Project Editor to take the fix list into the AI editor on your project page. The fixes appear as a checklist in the editor panel under Scan fixes. Select the ones you want to apply — individually or all at once — and click Apply fix or Apply N fixes to let the AI make the changes.
Run the scan from PRD Workshop
If you are working in the PRD Workshop (/prd-workshop), the scan is available at the choice screen after you paste or load a PRD. Choose Scan first (marked Recommended) to run the analysis before starting an edit session. After the scan completes you can continue directly into Workshop chat with the fix list pre-loaded as context.
Scan before you export
When you click Export on a project, Draftlytic may prompt you to scan first. You can choose Scan with AI to run the review before your file downloads, or Export anyway to skip it. Check Don't show this again to suppress the prompt on future exports.
Tips
- Set a target tool in your profile before scanning. The feedback is noticeably sharper when the AI knows whether you are targeting Cursor, Lovable, Bolt, or something else — it flags the specific gaps those tools struggle with.
- Don't close the dialog while the scan is running. Credits are deducted at the start of the call, and closing mid-scan does not trigger a refund.
- An Excellent score with zero fixes is a real result — the backend enforces this: if the AI returns zero fixes it always marks the project Excellent regardless of what else it said.
- You can apply fixes one at a time or batch them. Batching up to 8 fixes in a single AI edit call is more credit-efficient than running each fix separately.
- If you are on Starter and the scan surfaces a depth upgrade suggestion, it means your project was generated at Brief or Standard depth and the fixes involve acceptance criteria or navigation endpoints that Detailed depth fills in automatically on generation. Upgrade to Plus or Pro to unlock Detailed depth.
- Credits are refunded automatically if the scan times out or the AI returns an unusable response. You will see a toast message confirming the refund.
Credits & plan
PRD Scan costs 30 credits per run and is available on all paid plans (Starter, Plus, and Pro). The free tier cannot access the scan — clicking Scan with AI on a free account redirects to the pricing page.
Related guides
FAQ
What does the scan actually check?
The AI reads your full spec and looks for gaps (missing sections or features), vague requirements that an AI coding tool would interpret unpredictably, features with no assigned product tier, and — depending on how your project was generated — missing acceptance criteria or navigation endpoints. It returns up to 10 concrete fixes, prioritized by impact.
Are credits refunded if the scan fails?
Yes. If the scan times out, hits a token limit, or the AI returns an unusable response, your 30 credits are refunded automatically. The one exception is if you close the dialog while the scan is already running — credits are not refunded in that case because the AI call is already in progress.
What is the difference between a project scan and a PRD Workshop scan?
Both use the same backend and cost the same 30 credits. A project scan reads your structured Draftlytic project and can also surface a depth upgrade suggestion (shown to Starter users only, since Plus and Pro already have Detailed depth). A PRD Workshop scan reads the free-text PRD you are editing in the Workshop and feeds the fix list directly into the Workshop chat session.
Can I apply only some of the suggested fixes?
Yes. After clicking Open in Project Editor, the fixes appear as a checklist in the AI editor panel. You can select any subset and click Apply fix or Apply N fixes. Fixes you skip stay in the list until you dismiss them or run a fresh scan.
What does the depth upgrade suggestion mean?
If you are on the Starter plan, your project was generated at Brief or Standard depth, and the scan flagged missing acceptance criteria or navigation endpoints, you will see a callout explaining that Detailed depth generates those fields automatically. Detailed depth is available on Plus and Pro plans. If you are already on Plus or Pro, the callout is not shown because Detailed depth is already available to you.
Does the scan modify my project directly?
No. The scan only diagnoses — it scores the spec and lists fixes, but does not change anything. Modifications only happen when you explicitly choose fixes in the AI editor and click Apply.