Web Search API Now Available for Telnyx AI Agents

Web Search API is now available. AI agents and internal services can now perform web search, retrieve page content, and run cited research through a single Telnyx-owned API, authenticated with their existing Telnyx API key. No separate search provider account, no second API key to manage. $5 per 1,000 calls, billed through your existing Telnyx account.

What's new

• Web search endpoint (POST /v1/web_search): Real-time web search returning structured results with title, URL, description, and snippets. Supports count (1-100), country, safesearch, freshness, domain include/exclude filters, and livecrawl mode.
• Content retrieval endpoint (POST /v1/contents): Fetch clean HTML or Markdown from up to 20 URLs per call. Returns page title, HTML, Markdown, and metadata. Configurable crawl timeout and max-age cache control.
• Research endpoint (POST /v1/research): Deep research with cited answers. Runs multiple searches, reads sources, and synthesizes a response with inline citations. Three effort levels: lite, standard, deep. Supports background mode for long-running tasks with async polling via GET /v1/research/:task_id.
• Telnyx API key authentication: Every endpoint uses the same Bearer token you already use for Voice, Messaging, and Inference. No new credential to provision or rotate.
• Provider abstraction: The API exposes a stable, Telnyx-owned interface. Additional search providers can be wired in behind the same endpoints without any client-side change.

Why it matters

• Agents that need real-time web intelligence today require a separate search provider account, a separate API key, and a separate billing relationship. That is one more vendor boundary, one more credential to leak, one more integration to maintain when the provider changes their API.
• Every agent on the Telnyx platform already has a Telnyx API key. Web search, content retrieval, and research are now available through that same key, on the same platform that already powers the agent's voice, inference, and telephony.
• The provider abstraction means the search backend can evolve without breaking any client. Agents call Telnyx. Telnyx manages the provider relationship, the credentials, the rate limits, and the response shape.

Pricing

Example use cases

• Voice AI agents that need to answer questions about current events, live data, or anything outside their training cutoff can call /v1/research mid-conversation and return a cited answer without leaving the Telnyx platform.
• Autonomous research agents can kick off a background research task, poll for completion, and receive a synthesized answer with inline citations and a source list.
• Internal tooling and cron jobs can fetch clean Markdown from known URLs via /v1/contents for competitive intelligence, pricing monitoring, or content audits without scraping infrastructure of their own.

Getting started

1. Get your Telnyx API key from Mission Control, or use an existing key.
2. Call any endpoint with Authorization: Bearer <your-key>:

curl -X POST https://web-search.telnyx.com/v1/web_search   -H "Authorization: Bearer $TELNYX_API_KEY"   -H "Content-Type: application/json"   -d '{"query":"latest voice AI frameworks","count":5,"freshness":"week"}'

3. For deep research with citations:

curl -X POST https://web-search.telnyx.com/v1/research   -H "Authorization: Bearer $TELNYX_API_KEY"   -H "Content-Type: application/json"   -d '{"query":"Compare RAG vs fine-tuning for domain-specific QA","research_effort":"deep","max_sources":10}'

4. For background research, add "background": true and poll GET /v1/research/:task_id until status is completed.

Learn more in the Web Search docs or on the product page.