SERP API for AI agents
and RAG pipelines.
Plug your agent into a live web. Fresh Google SERPs, News, Images and Maps — returned as clean JSON, ready for function-calling. No stale index, no scraper to maintain.
How agent products run on our API.
Three loops — RAG retrieval, tool calls, multimodal grounding — all on the same endpoint surface. Start with Search, layer News and Images as the agent grows.
Fresh retrieval for LLMs
Plug Serpyx into your RAG pipeline as the live retrieval layer. The agent issues a query, we return ranked Google results with snippets and links — clean, citation-ready, ready to chunk for vector storage or pass directly into context.
// powered by Google Search API
- Real-time SERPs (no stale index)
- Citation-ready link + snippet
- JSON shape stable across queries
Tool calls for AI agents
Expose Serpyx endpoints as tools to Claude, GPT or any function-calling agent. The model decides when to query the live web — Search, News, Images, Maps — and gets parsed JSON back without scraping glue.
// powered by Google Search · News · Maps · Images
- Function-calling friendly
- Stable JSON contracts
- Sub-2s latency
Multimodal grounding
Combine Google Search, Images and News in a single agent loop. Ground LLM responses with fresh visuals and current news — no separate scraper, no proxy stack to maintain.
// powered by Google News · Images
- Text + visual grounding
- Time-bound news for freshness
- Same JSON shape as Search
The endpoints your agent will call.
Four routes share one search pool. Most agents lead with Google Search, layer News for freshness, then Images and Maps for multimodal and local context.
Google Search API
The retrieval layer for live RAG. Real-time SERPs, citation-ready snippets, fast.
GET /v1/google-searchGoogle News API
Time-windowed news for fresh-context RAG. Story clusters, source metadata, language targeting.
GET /v1/google-newsGoogle Images API
Image SERPs for multimodal agents — thumbnails, sources, dimensions, license.
GET /v1/google-imagesGoogle Maps API
Local context for location-aware agents — places, reviews, hours, geocoordinates.
GET /v1/google-maps
GDPR compliant
We process only public web data — no PII storage, no user tracking. Full GDPR alignment with signed DPA available.
CCPA compliant
California Consumer Privacy Act ready. We never sell personal data and ship clear opt-out endpoints.
Public web data only
Our scrapers fetch the same SERPs you'd see in a browser — public, indexed, citation-friendly. No PII stored.
Indexes drift — Google's doesn't. Serpyx hits Google in real time, so your agent always sees what users see today. For knowledge that changes (prices, news, product specs, regulations) a SERP API is materially more accurate than any cached embedding store.
Define a function with the JSON schema of our `/v1/google-search` endpoint and let your agent call it. Claude, GPT and Gemini handle this natively. We return parsed JSON the model can read without preprocessing.
P50 is sub-2 seconds, P95 under 3 seconds. Fast enough for a single user-facing turn. For multi-step agents, batch endpoints can run several searches in parallel and return them as a single payload.
Yes. Every result ships with a stable URL, page title, and snippet — exactly the fields LLMs need for grounded citations. We also include published dates on News results and source metadata on Images.
Yes. Residential, mobile and datacenter proxies across 190+ countries, automatic captcha solving, exponential retry on failure. Your agent gets a JSON response or it doesn't get billed.
Ship in minutes. Not weeks.
Get an API key in 60 seconds. Pull your first Google SERP as structured JSON — no proxies, no captchas, no maintenance.