Software at the resolution where decisions are made.

Aji IQ

Dish-level intelligence for diners. Know what to order, not just where to eat.

• DISH-LEVEL SCORINGEvery menu item gets a 10–98 score, not the restaurant.
• THREE ENGINES, ONE NUMBERBayesian quality model, 128-dimension taste vector, crowd-signal layer.
• EXPLAINED IN PLAIN ENGLISHHigh-confidence scores ship with prose: what reviewers love, what to note.
• TASTE-MATCHEDTwo diners see the same dish scored differently, based on their flavor profile.
• GROUP DININGTogether: invite friends (app or not), vote on dishes in real time, finalize consensus.
• MULTI-SOURCE FUSIONReviews and signals from 8 platforms blended into one score.

Aji PRO

Operating intelligence for restaurants. See your dish-level reputation, your menu's revenue gaps, your local competitive position.

• LIVE DISH PULSEReal-time alerts as new reviews and check-ins land.
• FLAVOR DNAEight-dimension flavor breakdown per dish, with the ingredients reviewers actually mention.
• COMPETITOR SHADOWHead-to-head benchmarking against nearby restaurants by dish, not by stars.
• AI MENU ENGINEERReads bottom-scorers, top-scorers, and gaps. Recommends what to remove, what to add.
• SENTIMENT DECODERThe recurring themes — positive, negative, neutral — pulled from your reviews.
• MULTI-LOCATION ROLLUPCross-location aggregates, with consistency scoring for menu items shared across sites.

One identity.

One sign-in across every Aji app. One profile, one social graph, opt-in cross-app sharing.

Personalized everywhere.

A 128-dimension flavor vector that learns from what you actually like. Works on dishes, drinks, and places.

One AI companion.

The AI shared across every app. Grounded in verified places, refuses to make things up.

The data underneath.

371,000 restaurants. 15.5 million menu items. Updated weekly. Behind every score, every recommendation, every plan.

Three-level priors.

A three-level prior hierarchy — global, cuisine, restaurant — that lets a new dish at a strong restaurant inherit confidence instead of regressing to a global mean. High-confidence scores require 30 mentions across two or more sources.

Personalized per diner.

Every diner has a learned flavor vector. Every dish has one too. Cosine similarity becomes “craving match.” Two diners standing in front of the same menu see different scores — accurately.

Grows with use.

First-party check-ins, ratings, and photos from Aji IQ users feed a live signal layer that grows with adoption. Weighting shifts as confidence grows: cold start leans on taste, mature dishes lean on quality + crowd.