Feedback Report Examples
See what ProofInk.AI produces. The two example feedback reports below were generated by the production system on a publicly-available research paper. They illustrate the difference between Quick Feedback and Deep Analysis, and show how revision-aware comparison works when you run a new feedback report against a previous one.
The Paper Being Reviewed
Both example reports analyze the same paper:
Konstantinos Spiliopoulos, "Information Geometry for Approximate Bayesian Computation,"
SIAM/ASA Journal on Uncertainty Quantification, Vol. 8, Issue 1, pp. 229–260 (2020).
The example reports were generated against the arXiv preprint
version (arXiv:1812.02127v2, August 2019).
Example 1 — Quick Feedback (v1)
Mode: Quick Feedback · Score: 4/5 · Processing time: 70.26 seconds
The Quick mode assessment of the paper is focused on the highest-signal issues a reviewer would flag first. It identifies 3 key strengths, 5 major weaknesses tagged with explicit page references, a few cross-cutting gaps and inconsistencies, and a 3-item executive action plan. Quick reviews are designed to surface the issues most likely to block acceptance — not to produce a criterion-by-criterion analysis.
Example 2 — Deep Analysis with Revision-Aware Comparison (v2 vs v1)
Mode: Deep Analysis · Score: 3/5 · Adjusted confidence: 74% · Processing time: 7 minutes
The Deep mode runs a considerably more sophisticated and involved multi-pass analysis. It produces an “At a Glance” 1-page summary with the main findings, and then a full criterion-by-criterion breakdown across ten dimensions (Hypothesis, Problem Statement, Methodology, Results, Novelty, Contributions, Limitations, References, Writing Quality, Discussion), each with its own score and detailed strengths, weaknesses, and suggestions. The report also includes an explicit cross-validation summary showing what the adversarial pass corrected or added.
Revision-aware comparison: this deep review was run with the earlier Quick review (v1) selected as the previous version. The resulting report includes a dedicated Revision Progress Dashboard comparing findings across versions: which issues were resolved, which remain open, which are new in the deep pass, and what progress the feedback detects overall.
Note on the "?" entries in Per-Criterion Changes: in this example, the previous version was a Quick review, which does not produce criterion-level scores — it produces an overall assessment only. That is why the dashboard shows "? → 4" for each criterion: the new Deep review has a score for Hypothesis (4), Methodology (3), etc., but there is no earlier criterion-level score to compare against. When you run a Deep review against a previous Deep review, you see the actual score progression (e.g., "3 → 4") for each criterion.
You can also see that the Deep mode surfaces substantially more strengths and more weaknesses than the Quick mode, with deeper technical detail, more page-specific citations, and a complete mapping of issues across the revision cycle.
Quick vs Deep — Which should I use?
Both examples analyze the same paper, but the two modes serve different purposes:
- Quick (~1–3 minutes): best for early drafts where you want a fast triage of the biggest issues before you invest time in a full revision. Ideal when you are still shaping the overall argument and want to catch structural problems cheaply.
- Deep (~5–15 minutes): best for a near-final draft where you want a criterion-by-criterion evaluation matching what a real review panel would produce. Ideal before submission, or when responding to reviewer comments on a revision.
A common workflow among beta testers: run Quick early, revise, then run Deep on the revised version with the Quick selected as the previous version. The Deep report then includes the revision-aware progress dashboard, showing exactly which of the Quick's concerns the revision addressed. Then, revise the manuscript/document based on the Deep report's findings and, if you want, run another Deep including the revision-aware progress dashboard against the previous Deep feedback report to see what was successfully addressed and what not.
What about grant proposals?
ProofInk.AI supports both Quick and Deep feedback for grant proposals, evaluated against agency-specific criteria from NSF, NIH, DOE, ERC, UKRI, Wellcome Trust, Simons Foundation, NSERC, CIHR, SSHRC, and others. Because grant proposals are confidential by their nature, we do not publish example grant reports publicly. Sign up for an account to see the grant feedback format in your own private workspace.
What about the Revision Response tool?
ProofInk.AI also helps you respond to reviewers after you have received their comments on a submitted grant proposal or paper and have prepared a revised version, by drafting a point-by-point response letter. Because reviewers' comments are oftentimes confidential, we do not publish example revision responses publicly. Sign up for an account to see the revision response format in your own private workspace.