VerdictPal
Join
GuidesPlaybook

Specific Aims one-pager

A submission-ready Specific Aims page that names the gap, three aims, the keystone, and one risk per aim.

Intermediate
Start now
Playbook

Specific Aims one-pager

A submission-ready Specific Aims page that names the gap, three aims, the keystone, and one risk per aim.

Steps5Time~140 minArtifactSpecific Aims one-pagerLevelintermediate
Before you start
  • A confirmed funder, call, and page limit
  • A working bibliography (Zotero, Mendeley, or a clean PDF folder)
  • A rough hypothesis you can defend without AI help
Tools in this guide
ClaudeNotebookLM
0%0/5 steps

PhD / postdoc grant author · 2.5–3 hours

Reviewers decide in 90 seconds whether your Aims page is worth a real read.

This playbook makes those 90 seconds defensible: gap, aims, keystone, risks — every line backed by a source and a reviewer-grade verification.

Specific Aims one-pager

Specific Aims one-pager

Single page: 2–3 sentence gap paragraph, three named aims, one keystone callout, one risk-and-mitigation line per aim.

Aim 2 (keystone): Test whether <X> mediates <Y> in <population>. Risk: small effect in pilot (n=24). Mitigation: powered for d=0.4 with preregistered stop rule.

Anchor the call

20 minClaude

Open the funder's call. Paste the verbatim review criteria, page limit, and formatting rules into Claude as system context before you write anything.

A 'call lock' note in your draft listing every named criterion you'll be graded against.
You are reviewing my Specific Aims against these verbatim criteria from <funder>. Repeat the criteria back to me as a numbered checklist — do not paraphrase. Then ask me three sharp questions before I draft.
Pitfall
Letting Claude paraphrase the criteria. Paraphrase is the bug, not the feature.
Verification
Every aim you draft can be mapped to a numbered criterion from the lock note.

Draft the gap in your voice

30 minClaude

Write a 2–3 sentence gap paragraph yourself. State what the field knows, what it doesn't, and why this matters now. Only after you have a draft, ask Claude to compress it to 90 words.

A gap paragraph that reads in your voice and survives a 90-word limit.
Compress this gap paragraph to 90 words. Preserve every claim and every source citation. Do not invent. If you remove anything, list what you cut underneath.
Pitfall
Accepting smoother prose that quietly drops your strongest claim.
Verification
A peer reading the compressed paragraph aloud can point to your specific contribution.

Write three aims, name the keystone

35 minClaude

Write each aim as a single sentence: verb-first, measurable, scoped to one method. Mark exactly one as keystone — the one that, if it fails, kills the project.

Three numbered aims plus a one-line keystone callout under Aim 2 (or wherever it sits).
Critique these three aims as a study-section reviewer. Which one is the keystone and why? Which one could be cut without weakening the proposal? Be specific, not encouraging.
Pitfall
Letting all three aims sound equally important. Reviewers grade hierarchy.
Verification
You can defend in one sentence why the keystone is the keystone.

Pressure-test against your literature

30 minNotebookLM

Load your bibliography PDFs into NotebookLM. Ask: which aim is most threatened by which prior work, and what is the strongest counter-argument from my own corpus?

One 'risk and mitigation' line per aim, each pointing to a real source in the corpus.
Pitfall
Trusting a grounded summary instead of opening the threatening paper.
Verification
Each risk line cites a paper you have personally opened in the last 24 hours.

Hostile-reviewer the final page

25 minClaude

Paste the full one-pager into Claude and ask it to roleplay a hostile study-section reviewer. Capture every objection that lands; address each in the doc or log a 'will not change' reason.

A red-team log: every objection labelled 'addressed' with the edit, or 'declined' with the reason.
Roleplay a skeptical NIH study-section reviewer. Be specific and mean — point to phrases, not vibes. List every weakness in order: framing, novelty, feasibility, rigor, significance. No reassurance.
Pitfall
Asking the model to be 'helpful' or 'constructive'. You want adversarial, not nice.
Verification
Every red-team objection is either reflected in a doc edit or logged with a defensible reason.
Finish line

You should now have: Specific Aims one-pager.

  • Every line on the one-pager maps to a named review criterion from the call lock.
  • Exactly one aim is marked keystone, and you can defend that choice in one sentence.
  • Each aim has one risk-and-mitigation line that points to a paper in your bibliography.
  • The red-team log lists every adversarial objection as addressed or explicitly declined.

PrivacyNever upload unpublished participant data, confidential clinical records, or unfiled IP into Claude, NotebookLM, or ChatGPT. The bibliography PDFs you feed NotebookLM should be published or licensed; the funder criteria you paste into Claude are public. If you cannot publish what you pasted, do not paste it.