Grounded Research API — Any Question, Your JSON Schema

Give it a question and the JSON schema you want back, and get a cited, live-Google-grounded, structured answer matching your schema. Built for developers and AI agents that need a fact, not a scraped page — no RAG pipeline to write, no page-fetching or synthesis step of your own.

Why this instead of a generic web-search/RAG tool

Most "AI web search" tools (including Apify's own RAG Web Browser) hand you raw scraped page text and leave the synthesis to you — you still have to parse multiple pages, extract the fact you actually want, and shape it into your own data model. This Actor does that whole pipeline for you: you describe the shape of the answer you want, and you get back exactly that shape, already populated and grounded in a live search, with sources attached.

Tutorial: how to run it

1. Click Try for free (or Start) on this page.
2. Fill in question and outputSchema — or leave the prefilled population example to see it work immediately.
3. Add context if the question needs disambiguating.
4. Click Start and open the Output tab once the run finishes.
5. Your requested fields come back populated, alongside sources and groundingUsed so you can verify how it was answered.

Common questions, answered directly

Is there an API that returns structured, cited answers instead of raw scraped text?

Yes — that's the whole point of this Actor: you describe the shape of the answer you want as a JSON schema, and you get back exactly that shape, populated and grounded in a live search.

How do I get an AI agent to look up a fact with a custom output schema?

Call this Actor with your question and outputSchema; no need to write your own prompt-and-parse logic for each new fact type.

What's an alternative to building my own RAG pipeline for simple fact lookups?

This — it replaces the "search, fetch pages, extract the fact, shape the data" pipeline most teams build by hand with a single API call.

What you can use this for

• AI agents / LLM apps that need a quick, cited fact lookup as a tool call, without building their own search+extract pipeline.
• Data enrichment pipelines — turn a list of entities into structured facts about each one, one schema, repeated calls.
• Chatbots and internal tools that need current, sourced answers rather than a model's stale training-data guess.

Input

Field	Description	Default
question	What you want researched	population example
outputSchema	JSON object describing the fields you want back, e.g. {"population": "number", "foundedYear": "number or null"}	example schema
context	Extra detail to scope/disambiguate the question (optional)	—
model	Gemini model	gemini-3.8-flash
apiKey	Your own Gemini API key (optional — leave blank to use the Actor's key)	—

Every field has a working default — hit Start with no changes and you'll get a result.

Output

Whatever fields you asked for in outputSchema, populated and grounded, plus:

{
  "population": 979882,
  "foundedYear": 1839,
  "state": "Texas",
  "sources": [{ "url": "https://...", "title": "..." }],
  "searchQueries": ["..."],
  "groundingUsed": true,
  "model": "gemini-3.8-flash",
  "verifiedAt": "2026-09-08T12:00:00.000Z"
}

Pricing

Pay-per-event: $0.15 per question researched. No subscription, no minimum.

Actor recommendations

• 🏢 Company & Person Verification API — a purpose-built version of this pipeline for company/person due diligence.
• 📈 AI Visibility Checker — a purpose-built version of this pipeline for AI-assistant brand mentions.
• 📊 B2B Sales Trigger Checker — a purpose-built version of this pipeline for company buying-signal research.