Agents & Grounded Content
Review Summarizer
Extracts sentiment, themes and recurring complaints, with every finding tied to review IDs.
The Problem
Hundreds of reviews hide the two or three issues that matter. Summaries need to be checkable.
How It Works
- Derive sentiment from ratings and split positive/negative reviews.
- Count themes with a keyword dictionary.
- Surface complaint phrases only when at least two negative reviews repeat them, with sample review IDs as citations.
Technically Interesting
Citations are built in: each finding links back to reviews you can read.
Why It Matters
A summary a product team can verify is a summary they can act on.
Scope & honesty. Keyword-dictionary analysis over 80 synthetic reviews; not a sentiment model.