Tastora

• 

  Same city. Same cuisine. Different person. A Frank Sinatra taste signal reshapes Qloo's top Italian picks in New York.

• 

  Measured, not claimed: Frank Sinatra brought in 6 new places and moved all 10 compared with a plain search.

• 

  Good food should still feel like home. Tell Tastora who you're choosing for, and watch their tastes become restaurant picks.

• 

  Every pick ships with receipts: Qloo's photo, one clear reason, needs sorted at a glance, and the exact questions to ask.

• 

  Rules Tastora keeps, enforced in code. Allergies are never marked safe. Missing information always means "confirm".

• 

  Print the checklist and pick up the phone. Every question to ask the restaurant, in one place.

Inspiration

Food apps optimise for the crowd. Not for the person you're feeding.

When you're choosing for someone you care about, (a parent, friend, sibling, or grandparent) popularity means very little. What matters is what they love, what they can't eat, and whether a place will actually work for them.

Qloo's brief highlighted dining for older adults, and I took that personally. No age assumptions, No guesswork. Start with the person: the food they grew up with, the musician they still hum along to, and the things they must avoid.

What it does

Tell Tastora who you're choosing for. Their favourite foods, taste memories like musicians or films they love, and their non-negotiables: allergies, dietary requirements, soft-food needs, a quiet table, or wheelchair access.

Tastora finds restaurants nearby and gives you more than a list of names. Every recommendation comes with context whch are listed below.

• A restaurant photo sourced through Qloo.
• Why it matches their taste, with a clear explanation.
• Their needs, organised into matched, conflict, or needs confirmation.
• Dishes worth asking about, and dishes set aside with the reason why.
• Specific questions to ask when contacting the restaurant.

Allergies are always marked for confirmation. No app can guarantee a dish is safe without verifying its ingredients and preparation. Tastora doesn't pretend otherwise.

Share the shortlist with family on WhatsApp or print the checklist. Call the restaurant with the right questions, then decide.

How I built it

Next.js and TypeScript, deployed on Vercel.

An AI agent plans each search: which cuisines to explore, which taste memories Qloo recognises, and how to construct the search. Qloo provides the taste-based restaurant ranking. Tastora's own rules engine, backed by 341 tests, evaluates each recommendation against the person's requirements.

The separation is deliberate: AI decides how to search. Tested code handles constraints and safety. The AI never sees allergy or dietary data.

What makes it Qloo-powered

Qloo isn't an add-on. It powers the restaurant discovery and taste-based personalisation at the centre of Tastora.

We measure its impact by running each search twice: once with the person's taste signals and once without.

• Frank Sinatra fan, Italian, New York: 6 of the top 10 recommendations change.
• Bob Marley fan, Caribbean, the Bronx: 0 of 10 recommendations overlap with a plain search.

Same city, Same cuisine, Different person, Different table.

What's next

Real menu data. Expanded city coverage, verified market by market. Shared taste profiles that help families make better decisions together.

Built With

• accessibility
• ai-agents
• anthropic-api
• claude
• css-modules
• eslint
• github
• google-maps
• llm
• next.js
• node.js
• qloo
• qloo-insights-api
• react
• responsive-design
• streaming
• svg
• tool-use
• typescript
• vercel
• vitest
• web-share-api
• whatsapp
• zod