DocumentationAPI Reference
    Intelligent Agent Task Automation

    Intelligent Agent Task Automation

    Desearch gives AI agents fresh evidence before they write, plan, analyze, or act. Agents call Desearch APIs as tools for web search, social search, source discovery, and deeper research.

    Agent workflows

    • Research agent: search the web and X before drafting an answer. To focus on Reddit threads, call GET /web with site:reddit.com in the query.
    • Sales or market agent: monitor customer segments, competitor launches, and buying signals.
    • Social intelligence agent: retrieve X, TikTok, Facebook, and Instagram context before generating reports.
    • RAG pipeline: extract or search fresh sources before embedding or summarizing.

    Recommended tool mapping

    Agent needDesearch endpoint
    Fresh multi-source contextPOST /desearch/ai/search
    Source links onlyPOST /desearch/ai/search/links/web
    X/Twitter contextGET /twitter or X retrieval endpoints
    Web result listGET /web
    Full page textGET /web/extract

    Example agent tool call

    bash
    curl --request POST 'https://api.desearch.ai/desearch/ai/search' \ --header "Authorization: $DESEARCH_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ "prompt": "What changed this week in AI search benchmarks?", "tools": ["twitter"], "date_filter": "PAST_WEEK", "count": 10, "streaming": false }'

    Implementation tips

    • Add retries/backoff for 429 and transient 500 responses.
    • Treat 422 as a prompt/request validation bug and surface a useful developer error.
    • Record X-Desearch-Service, X-Desearch-Usage-Count, and X-Desearch-Cost-Usd in agent logs.
    • Never expose API keys in agent-generated artifacts or browser code.