Reply to App Reviews From a Screenshot

Most review-reply workflows assume you're already inside a dashboard, looking at a review, with the right app selected. In practice, that's rarely how a review actually reaches you. Someone on your team screenshots a 1-star review from Slack, or forwards one from a support ticket, and now there's a gap between "I saw this" and "I replied to this."

ReviewDesk's screenshot-to-reply agent closes that gap. Paste a screenshot into the agent chat, and it reads the text with Gemini Vision OCR, fuzzy-matches it against your actual review database using pg_trgm, and comes back with a confidence-scored match and a drafted reply — without you ever opening the dashboard.

How the matching actually works

A screenshot is messy: cropped edges, a partial username, a review truncated mid-sentence. OCR gets you a rough transcript, not a clean match. The agent runs that transcript through a fuzzy trigram match against your stored reviews rather than requiring an exact string match, so a screenshot that's missing the last line of a review — or has an OCR typo in the reviewer's name — still resolves to the right review with a confidence score attached.

If the match confidence is low, the agent says so, rather than confidently drafting a reply to the wrong review. That distinction matters more than it sounds — a wrong-review reply posted to a live store page is exactly the kind of mistake that's hard to walk back.

Nothing posts without confirmation

Every draft generated this way lands in a pending_drafts state. The agent's response grammar is simple: send, edit <text>, or cancel. Nothing reaches the App Store or Google Play until you explicitly choose one of those three. That's the same confirm-before-post principle that runs through every automated path in ReviewDesk — AI speed without AI accidents on a page your users can see.

Where this fits into a real workflow

• A teammate forwards a review screenshot in Slack — paste it into the agent instead of hunting for it in the dashboard.
• Support flags a review during a call — screenshot it, get a draft back in seconds, confirm from the same chat.
• You're triaging reviews from your phone and a full dashboard isn't practical — the same flow works from any chat surface the agent is connected to.

It's a small feature on paper, but it removes the one step that quietly kills a lot of review-reply intentions: the gap between noticing a review and being in the right screen to answer it.