Mealio understands a craving the way a friend does — and answers with places that provably serve it.
The dish, the price, the vibe, the block. Ranked by relevance, never by who paid.
Built like a search engine. Reads like a friend.
The whole stack is ours — from understanding the sentence to the corpus it searches to the ranking that answers. Nothing is fetched from a third-party search API at runtime.
One of the largest structured restaurant datasets anywhere
Purpose-built pipelines turn menus, hours, happy hours, reservations, delivery, accessibility and photos into clean, queryable structure before the algorithm ever touches it.
- 2B+ data points — structured fields and machine-applied tags
- 1.9M menu items, structured down to the dish and its price
- Compounding daily — every day the engine runs, the corpus deepens
From spoken thought to proven results
The engine understands structure, not keywords — and it shows its work.
- Full-sentence decomposition — one thought, phrased however it comes out, understood as everything it asks for at once
- Nuance survives the ranking — it knows a shrimp cocktail from a cocktail
- Evidence-bound answers — every match ships the menu items that satisfy the ask
Speed is a feature we refuse to trade
Depth usually costs speed. Engineering that trade away is the product.
- ~2s end to end — query understood, corpus ranked, receipts attached
- <150ms core ranking pass across 20K+ venues
- Live, computed results on every query — never a canned list
Every change runs the gauntlet
End-to-end suites of real cravings, scored by an independent judge, plus adversarial red-team sweeps. Every solved edge case becomes a permanent regression test — the engine never relearns an old lesson.
94% less AI than the next best search engines. And better.
AI is a great tool for building new things. Like all technology, it should be used responsibly and with humanity in mind. We hold our AI usage at 94% less than Google Maps, Google Search or ChatGPT — while being a better restaurant search engine than any of them.
Most of live search is re-analyzing the same data over and over — re-reading reviews, reconciling menus, chasing SEO. The data that actually decides a match only needs to be captured periodically. Our database is built specifically for restaurants, so a model never spends tokens working out what you are asking about. The structure already knows.
Across approximately 15,000 test searches, producing a comparable pool of results with the supporting evidence took roughly 17× the tokens for ChatGPT and general search engines — for less accurate results, a narrower pool, and one polluted by SEO and pay-to-rank. The restraint is where the quality comes from.
What only Mealio does.
| Mealio | ChatGPT | OpenTable / Resy | Delivery apps | Beli | ||
|---|---|---|---|---|---|---|
| Full-sentence cravings understood | ✓ | ~ | ✓ | — | — | — |
| Proof on every match (menu receipts) | ✓ | — | — | — | — | — |
| Menu data structured for search | ✓ | ~ | — | — | ~ | — |
| Never pay-to-rank, no ads | ✓ | — | ✓ | — | — | ✓ |
| Live results in seconds, every time | ✓ | ✓ | — | ✓ | ✓ | ✓ |
| Block-level precision (“8th Ave between 42nd & 50th”) | ✓ | ~ | — | — | — | — |
| Budget enforced against real menu prices | ✓ | — | — | — | — | — |
| Quality graded beyond star averages | ✓ | — | — | — | — | ~ |
Capability comparison as of August 2026, from publicly observable product behavior.