Wine intelligence for AI agents

The first wine MCP server and REST API built for LLMs. Every recommendation backed by verified competition medal data — not hallucinated tasting notes.

116,775
Wines
124,983
Competition Medals
16,302
Producers
65
Countries
207
Competitions
Install MCP Server API Documentation

Why AgentSomm?

LLMs hallucinate wine

They invent vintages, fabricate tasting notes, recommend wines that don't exist. AgentSomm gives agents real data to work with.

Medal data is the moat

124,983 verified competition medals from 207 international wine competitions. No other agent tool has this.

Agent-native interface

MCP server for Claude/ChatGPT, REST API for LangChain/CrewAI. Structured JSON responses agents can reason about.

How it works

// Your user asks their AI assistant:
"Find me a bold red wine for steak night, under $30, something award-winning"
// The agent calls AgentSomm:
GET /api/recommend?preferences=bold+red&food_pairing=steak&budget=under+%2430

{
  "recommendations": [{
    "wine": "Earthquake Zinfandel 2022",
    "producer": "Michael David Winery",
    "region": "Lodi, CA",
    "price": 18.99,
    "awards": [
      {"competition": "SF Chronicle", "medal": "Double Gold"},
      {"competition": "TEXSOM", "medal": "Gold"}
    ],
    "why": "Multi-award Zinfandel with bold dark fruit — classic steak pairing at $19"
  }]
}

MCP Tools

recommend_wine

Get wine recommendations by preferences, budget, occasion, food pairing

search_wines

Search by name, varietal, region, price, or medal level

wine_details

Full profile with competition medal history

compare_wines

Side-by-side comparison of two or more wines

food_pairing

Wine suggestions for a specific dish