I built this because I couldn't stop thinking about it.

For years I watched restaurants and small hotels around me run on a strange contradiction. Every guest who walked through the door left a paper trail — a Google review, a TripAdvisor write-up, a sentence about the lamb, a complaint about the waiter — and the operators almost never read any of it.

This is exactly what a machine should do. Read everything, in any language, every day. Surface the actual patterns. Tell the operator the one paragraph they need before service starts. Not a dashboard with twelve charts — a friend who already did the reading.

Late nights, then weekends, then full-time. Piece by piece:

• 01The scraper that watches every new Google and TripAdvisor review as it lands.
• 02The language models that score sentiment across six aspects and translate any of 30+ languages.
• 03The narrative brief that reads like a human wrote it — what moved, what to fix, what to celebrate, three actions.
• 04The pricing — rewritten three times until it felt fair. One flat rate per venue, another small one per competitor, nothing else.
• 05The onboarding — paste a URL, get the analysis two minutes later.

Verdscore — an AI that reads every review of your restaurant, tells you what changed, drafts the replies, and warns you the moment something moves. It runs while you're running service.

The signal is in the reviews. The job isn't reading them — it's acting on them. AI should compress the noise into a takeaway, not add another dashboard to babysit.