Skip to content
In this sectionWrite the proposal

Get an AI review

An AI review reads every drafted section against the funder's scoring criteria, or standard criteria when the funder doesn't publish any. It scores each one and suggests what to fix before a person reads the draft.

  • 5 min
  • Every plan
  • Uses 1 AI action

Screens from October 2026

In the app:Pipeline, then Proposal, then Review

Run a review

Before the first review, the tab explains what a review does. What will be reviewed lists each section with its length, and the budget explanation if there is one. Sections without text show Skipped · empty, since the review reads only what's written.

The Review tab before a first review: what a review does, Review our draft, and What will be reviewed, with each section's length and the standard criteria.
The Review tab of a proposal that hasn't had a review yet. Example workspace · fictional data

Select Review our draft. A review takes up to a minute and uses one AI action. Nothing is sent to the funder. If you're applying as an individual, the button reads Review my draft. After the first review, it reads Review again.

Which criteria it uses

If GrantFlex has summarized the funder's guidelines and found scoring criteria, the review uses them, with their points. Otherwise it uses standard criteria: need, project plan, ability to deliver and budget. The box under What will be reviewed says which, as Criteria: from the funder’s guidelines or Criteria: standard.

Each criterion is scored out of the funder's points, or out of 10 when the funder publishes none. When every criterion has points, the review adds them up, such as 22 / 40 pts.

Read the scoresheet

The scoresheet starts with a verdict: Looks submission-ready, Needs work or Not ready. A short summary says why, and the total score follows.

The verdict of the Arts Access review: Needs work, a summary of what holds the draft back, and 22 of 40 points.
The verdict on Lanternfield's draft, reviewed against the standard criteria. Example workspace · fictional data

Below the verdict, each criterion has its score, a critique of what the draft says or leaves out, and fixes. Each fix names the section to change. At the end, Not addressed anywhere lists what no section covers.

The Program Design criterion scored 4 out of 10, with the critique and three fixes, each with Revise with this suggestion and Open section.
Example workspace · fictional data

If the proposal is long, the review may leave some sections out to fit. A note at the top names them, so you can check those yourself.

Check first

The score is guidance from AI, not the funder's decision, and it can vary between reviews. Check each suggestion against the funder's guidelines before you change the draft.

Revise with a suggestion

Beside each fix, Open section opens the section so you can edit it yourself. Revise with this suggestion has AI rewrite the section to apply that one fix. It's offered when the fix names one of your sections, that section has text and the proposal isn't locked.

  • A dialog shows the fix, then the current section and the revision with the changes marked, and the new length against the limit.
  • Fill any [ADD: …] placeholders the revision adds before you approve the section.
  • Select Use this revision to save it, or Keep current text. Edit the section yourself opens the section instead.

Each revision uses one AI action. The text it replaces stays in the section's version history. On the Arts Access proposal, Program Design's schedule fix was applied this way, and its history shows the change as AI revision accepted.

Review history

Every review is kept. Once there are two or more, Review history lists each run with its date and verdict, so you can see whether your changes moved it. Select a run to see its scoresheet again.

Next

Suggested