Skip to main content

Overview

Agentic Review lets you define review programs — reusable configurations that describe how the AI should review documents — and execute review runs against target documents. The AI analyzes each document and produces structured action items for human review.
Review Inbox with review runs table

Core Concepts

Review Program

A reusable template with review instructions and reference guideline files.

Review Run

A single execution of a program against selected documents, producing action items.

Action Items

Specific findings with title, description, location, and severity.

Creating a Review Program

1

Write the review prompt

Describe what the AI should look for — compliance gaps, formatting issues, missing sections, etc.
2

Attach guideline files

Select reference documents from your workspace that the AI should use as review criteria.
3

Save the program

Programs are reusable — run them against different documents as many times as needed.

Running a Review

  1. Select a review program
  2. Choose the target documents to review
  3. Optionally set a due date
  4. Start the review — the AI runs autonomously
The AI agent reads your target documents, compares them against guidelines, and creates action items for any issues found.

Action Items

Each action item includes:
FieldDescription
Title5-10 word summary of the finding
Description1-2 sentence explanation
LocationPage number and section reference
Severityinfo, warning, or error
Action items can be individually resolved or unresolved as your team addresses them.

Disposition Codes

After the AI completes its review, a human reviewer assigns a final disposition code. Raycaster Doc supports multiple presets:
PresetCodesUse Case
EPCA, B, C, DConstruction document review
Life SciencesApproved, Conditional, Revise, RejectedPharma/biotech QA
BLA SubmissionBLA Ready, Minor Deficiency, Major Deficiency, Refuse to FileFDA regulatory
SimpleApproved, Changes, RejectedGeneral review

Re-running Reviews

Reviews can be re-run against the same or updated documents. Follow-up reviews have access to the previous review’s findings for context, enabling iterative improvement.