Welcome to Agentic Feedback Synthesis thatreads every summary statement all the way through.

Reviewer feedback is the most expensive advice an institution receives and the least used. Risely's agents read every summary statement in full, extract what actually moved the score, and hand the PI a resubmission plan.

Synthesis · one PI's resubmission

reading the summary statement

Aim feasibility drew both critiques

address first

reviewers 1 and 3 flagged the same missing pilot data · the new preliminary figure answers it

The rigor section scored well

protect

keep the structure · a revision must not rewrite its own strengths

Institution-wide pattern

systemic

the same power-analysis critique has hit six proposals this year · a template fix is drafted for the office

The plan assembles once the reading is done.

What moved the score gets named.

A summary statement gets read once in anger and filed. The synthesis reads it cold, weighs each critique by who raised it and how hard, and puts the score's real anchor at the top of the plan.

Critiques ranked by weight

Missing pilot data behind one aim

both reviewers · the score's anchor

A thin power analysis

one reviewer · fixable in a page

A quibble about a citation

cosmetic · answer in one line

A summary statement read once in anger treats every critique as equal. The synthesis weighs them.

What scored well gets protected.

The quiet failure mode of resubmissions is rewriting the sections reviewers liked. Everything that scored well is marked protect, so the revision cannot lose ground by accident.

The revision boundary

Approach, first aimrevise · answers the pilot-data critique
Rigor and reproducibilityprotect · scored well as written
Team and environmentprotect · praised by both reviewers

Resubmissions usually lose ground by rewriting their own strengths. The protect marks make that hard to do by accident.

The pattern across the portfolio surfaces.

The same power-analysis critique has hit six proposals this year, and nobody read across them until now. One template fix, drafted for the office, lifts every submission that follows.

One critique, six proposals

reading across

·
·
·
·
·
·

six proposals this year · the same power-analysis critique

Nobody reads across proposals on stolen time. The synthesis does, and one methods-template fix lifts every future submission at once.

The response map writes itself.

Every response in the resubmission points at the exact comment it answers, in the reviewer's own words. The study section sees that it was heard, which is the strongest signal a resubmission can send.

Every comment gets a response

Reviewer 1 · pilot data missing

new preliminary figure, first page of the response

Reviewer 3 · the same gap

the same figure answers it, in their words

Reviewer 2 · power analysis

expands to one page, per the template fix

Answering a reviewer's exact words is the strongest signal a study section can receive. The map makes it the default.

One statement lands. The synthesis re-runs.

A new summary statement, a program officer's note, a declined resubmission. Any of them adds to what the portfolio knows, and every one runs the same way.

  1. A summary statement arrives

    read in full the day it lands

  2. The critiques cluster

    two reviewers, one underlying gap

  3. The plan updates

    address, protect, leave alone

  4. The pattern check runs

    the same critique shows up across the portfolio

  5. The PI starts from a plan

    instead of a pile of PDFs

What changes, and how it runs.

Resubmissions answer the actual critique

The first paragraph responds to the exact comment.

Strengths survive the revision

What scored well is marked and left alone.

One fix lifts the whole portfolio

The shared critique becomes a template change for the office.

The technical read

Connects read-only to your grants system and review archive to start

The PI decides what the resubmission changes

Every action lands on an append-only log

SOC 2 Type IIFERPAGDPR

Runs alongside what you have. Nothing rips out.