Ramen Bar
Tonkotsu, 8 min wait
Place memory for agents
Palette
Give every agent a memory for places it has tried, reviewed, liked, skipped, photographed, and recommended.
Palette turns scattered food, venue, and local-service experiences into structured reviews agents can write now and search before the next call.
Corner Bakery
Quiet tables, strong Wi-Fi
Taco Window
Late night, fast line
Next best match
Ramen Bar 92% confidence
$ palette search --q ramen --near me
=> Ramen Bar · 5.0 · tonkotsu · short wait
$ palette post --entity-name "Cafe Nova" --rating 4
=> saved review · quiet tables · good for work
What it unlocks
A simple review brain for agents that need to record and recall local context.
Remember every place
After an experience, agents can write structured reviews with ratings, notes, media, and context.
Find the right answer later
Search by location, category, rating, keyword, or custom metadata when the next request comes in.
Make recommendations feel earned
Turn past visits and structured reviews into better suggestions for food, venues, services, and more.
Built for agent workflows
Agents get a clean CLI and JSON contract today, with room for shared backends later.
Under the hood
Consumer-simple on the outside. Agent-native underneath.
Palette runs locally, stores structured entities and reviews, and gives agents a dependable command surface for posting new reviews, searching old ones, and retrieving context.
$ palette post --entity-name "Ramen Bar" --rating 5 --text "great tonkotsu, short wait"$ palette search --near 37.77,-122.41 --q ramen --min-rating 4