recommend

Get personalized restaurant recommendations from a natural-language query. The AI considers occasion, preferences, location, and dimensional data to rank results.

Tier: Concierge ($80/1K calls)

Use cases

Power conversational dining recommendations in AI assistants
Match restaurants to specific occasions ("quiet anniversary dinner", "quick lunch near the office")
Replace multi-step search-and-filter workflows with a single natural-language query

Parameters

Name	Type	Required	Default	Description
`query`	string	Yes	--	Natural-language dining request (e.g., `"A quiet dinner for two in Mayfair, nice wine list"`)
`location`	string	No	--	City or area to bias the search toward
`latitude`	number	No	--	Latitude of search center (use with `longitude`)
`longitude`	number	No	--	Longitude of search center (use with `latitude`)
`limit`	number	No	5	Number of results (1-10)

Response

Field	Type	Possible values	Description
`status`	string	`"ok"`, `"error"`, `"no_coverage"`	Response status
`query_understood`	string	--	The AI's rephrasing of your query (confirms interpretation)
`location_used`	string	--	Resolved location used for the search
`total_candidates`	number	--	How many restaurants were considered
`results`	array	--	Ranked recommendations (see fields below)

All basic lookup fields (seemor_id, name, address, coordinates, cuisine_tags, price_level, grade, grade_label, tldr, neighborhood, city, seemor_url, coverage_level, analysis_date), plus:

Field	Type	Possible values	Description
`relevance`	string	`"strong_match"` (fit score >= 75), `"good_match"` (>= 55), `"match"` (< 55)	How well this restaurant fits the query
`match_reasons`	string[]	--	Why this restaurant fits (e.g., "Low noise level", "Strong wine list")
`distance_km`	number \| null	--	Distance from search coordinates. Null if no coordinates provided.

Example request

curl -X POST https://api.seemor.ai/api/mcp \
  -H "Authorization: Bearer sk_seemor_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "recommend",
      "arguments": {
        "query": "A quiet dinner for two in Mayfair, nice wine list",
        "location": "London",
        "limit": 3
      }
    }
  }'

Example response

{
  "status": "ok",
  "query_understood": "Quiet, intimate dinner for two in Mayfair with a strong wine selection",
  "location_used": "Mayfair, London",
  "total_candidates": 127,
  "results": [
    {
      "seemor_id": "...",
      "name": "The Square",
      "grade": "A",
      "grade_label": "Excellent",
      "price_level": "very_expensive",
      "tldr": "Refined modern European dining with an extensive wine program...",
      "relevance": "strong_match",
      "match_reasons": [
        "Low noise level - ideal for conversation",
        "Extensive, award-winning wine list",
        "Located in the heart of Mayfair"
      ],
      "distance_km": 0.2
    }
  ]
}

Notes

Response time is 3-8 seconds (AI-powered matching and ranking).
Location can be specified in the query text, as a separate location parameter, or via coordinates. If both query text and location parameter include a location, the location parameter takes priority.
Returns status: "no_coverage" if the requested area has insufficient restaurant data. This is not billed.
Requires a paid API key. Free keys receive a blocking error.