For AI agents & LLMs

Scout, for agents.
Company intelligence your AI can call.

Point Claude or any LLM at Scout and it can look up Nordic companies, read explainable scores and pull signal timelines — with machine-readable docs and MCP tools built in.

Machine-readable entry points

Everything an agent needs to self-serve. No human in the loop.

Connect via MCP

Add Scout as a set of tools in Claude (and any MCP client). A remote MCP server is rolling out — until then, point your agent at the OpenAPI spec, or use the cURL tools below.

claude_desktop_config.json
// Streamable HTTP MCP — add to Claude Desktop / Code (or any MCP client).
{
  "mcpServers": {
    "scout": {
      "url": "https://mcp.scout.jakvab.se",
      "headers": { "Authorization": "Bearer $SCOUT_KEY" }
    }
  }
}

The MCP server exposes search_companies, get_company and get_signals — thin wrappers over the REST API, using your API key.

Canonical tasks

The handful of calls that cover most agent jobs. Base URL https://api.scout.jakvab.se/v1, auth Authorization: Bearer <key>.

GET/companies/NO/923609016Look up a company by org number → entity + ai_summary + scores
GET/companies/NO/923609016/signalsIts signal timeline (new_ceo, hiring, funding…)
POST/search{"query":"logistics companies hiring in Oslo","limit":10}
GET/companies/{id}Look up by internal id (cmp_<COUNTRY><orgnr>)
example.sh
curl https://api.scout.jakvab.se/v1/companies/NO/923609016 \
  -H "Authorization: Bearer $SCOUT_KEY"

{
  "legal_name": "EQUINOR ASA",
  "ai_summary": "Norwegian energy company…",
  "scores": { "buying": 0.35 },
  "why": "based_on_signal_ids: […]"
}

Every score carries an explanation and based_on_signal_ids — cite them, don't guess. NO is live; SE/DK/FI are coming. Free tier: 10 requests/day, ≤100 objects/response (X-RateLimit-* headers).