UNSAID

Ask Claude for a St. Louis hidden gem: 13 times out of 16, it's Mai Lee. Tell UNSAID an artist, a show and a dish you love, and Qloo sends you where AI didn't.

Comment

UNSAID · AI sends everyone to Mai Lee. Your taste sends you elsewhere.

Ask Claude for a hidden gem in St. Louis and it says Mai Lee, on 13 of 16 tries. ChatGPT says it 10 times. Half of everything Claude recommends in the city is the same 29 places. As agents become how people decide where to go, everyone gets sent through the same few doors. The rest of the city goes unsaid.

UNSAID goes the other way. Tell it an artist, a show and a dish you love: say Frankie Beverly & Maze, The Original Kings of Comedy and rib tips. It crosses out the places the AIs named in over 900 answers about the city, Mai Lee included, and Qloo ranks what's left for your taste. You get seven places, each rated 4.3 or better, with a photo and why: because you love rib tips.

It holds on questions it never saw. On 40 real Reddit questions we never used to build the list, the AIs named 0 of UNSAID's 140 picks, against 32 of Qloo's plain top picks. And they aren't just obscure: UNSAID's fit your taste at the 75th percentile, random leftovers at the 47th.

Agents, but with taste. UNSAID is also an MCP tool, so your agent hands you your seven instead of the list everyone gets.

Try it in 60 seconds

  1. Open https://unsaid.mi11ion.io and tap Frankie Beverly & Maze · The Original Kings of Comedy · rib tips.
  2. Type three things you love. Your seven arrive in about half a minute.
  3. Paste any AI's St. Louis answer under your seven. If it names one, that card is crossed out and a new one takes its place.
  4. Give it to your agent: claude mcp add --transport http unsaid https://unsaid.mi11ion.io/mcp

How it uses Qloo, request to result

  • Your loves: /search and tags resolve an artist, a film or show, a dish
  • The pool: /v2/insights, restaurants ∪ bars near you, per love; dishes through filter.tags
  • Your rank: each love rescores the same fixed pool (filter.results.entities)
  • The lift: the same pool rescored for 97 tastes drawn from Qloo's most popular entities, plus popularity; your seven are where you beat it most
  • The card: Qloo's photo, description, hours, price and what people order

Why Qloo: it knows what fans of your artist, show and dish go on to like; an LLM reaches for what's famous. The AIs only decide what's off the table.

Who pays

Agents, per MCP call, for the one answer no other assistant gives. Tourism boards and business districts, per diner UNSAID sends somewhere AI didn't name. Restaurants never pay, so no spot is for sale.

Limits

Qloo predicts fit; it can't promise you'll love a given place. The crossed-out list is a dated snapshot of what Claude, ChatGPT and Gemini said, with and without web search, in St. Louis and New York; step 3 catches anything newer. Only your three loves leave your browser, to Qloo and, where available, one live Gemini check: no names, no accounts, keys stay on our server.

Built With

Share this project:

Updates

Submission history