surfaces: ["google_ai_overview"]AI Overviews sit above the organic results and answer the query before anyone scrolls. There is no official API for them. We render the real results page on our own browser fleet and return the AI Overview as a structured Envelope — the answer, the sources it links, and, critically, whether it appeared at all: absence is first-class data, not an error.
curl -sS -X POST "https://api.aisearchapi.dev/v1/search" \
-H "Authorization: Bearer $AISEARCH_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"query": "how to fix a leaking tap",
"surfaces": ["google_ai_overview"],
"regions": [{ "country": "US" }, { "country": "GB" }, { "country": "DE" }],
"method": "auto"
}'
# → 202 { "jobId": "…", "children": ["….google_ai_overview.…"] }
# then poll GET /v1/jobs/:childId — or add ?mode=sync for an inline EnvelopeOne canonical Envelope — the same four sections every surface returns — with the fields Google AI Overview actually populates. Field names below are the contract, not a summary.
provenance.surfacePresentThe field that matters most here: AI Overviews trigger on some queries and not others. false means the SERP rendered with no Overview — the job completes, flagged surface_absent, and the screenshot proves it.
answer.text · answer.markdown · answer.blocksThe AI Overview’s text as rendered in the block, normalized.
evidence.sources[]The pages the Overview links — the citation set GEO teams live on — with roles and cited booleans.
evidence.fanOutRelated queries observed around the Overview, provenance-marked.
provenance.regionRequested-vs-effective country. Overview presence and content vary hard by market — capture the same query across all nine v1 countries and diff.
job.artifactsScreenshot + proof-of-page HTML of the actual SERP, retained durably — evidence the Overview said what you claim it said.
{
"job": {
"id": "job_c2e84d17.google_ai_overview.own-fleet.US",
"surface": "google_ai_overview",
"method": "own-fleet",
"status": "completed",
"warnings": [],
"artifacts": {
"screenshotKey": "shots/2026/06/30/job_c2e84d17.png",
"proofHtmlKey": "proof/2026/06/30/job_c2e84d17.html",
"rawKey": "raw/2026/06/30/job_c2e84d17.json"
}
},
"provenance": {
"acquisition": "own-fleet",
"fidelity": "consumer_ui",
"surfacePresent": true,
"region": { "requested": "US", "effective": "US" }
},
"answer": { "text": "…", "markdown": "…", "blocks": [ … ] },
"evidence": {
"sources": [
{ "id": 0, "url": "https://…", "title": "…",
"role": "cited", "cited": true }
],
"fanOut": { "provenance": "observed", "queries": ["…"] }
}
}
// and when no Overview triggers — absence-is-data:
{
"job": { "status": "completed", "warnings": ["surface_absent"] },
"provenance": { "surfacePresent": false },
"answer": { "text": "", "markdown": "", "blocks": [] }
}The AI Overview block captured live via own-fleet browser render of the real results page. The managed-vendor SERP lane is phase-2; there is no official API.
Phase-2 lanes are roadmap, not product — pinning one returns an honest 422 today. The full picture is on the capability matrix.
By rendering the real Google results page in a stealth browser on our own fleet — the same page a user in that country sees — and extracting the AI Overview block. There is no official Google API for Overviews; anything that claims consumer-UI fidelity has to render the SERP.
That is a result, not an error. The job completes with surfacePresent:false and a surface_absent warning, you are charged nothing, and the screenshot and proof-of-page still exist — so "no Overview shown in DE on this date" is a provable data point. Overview trigger rates are exactly what visibility teams need to measure.
Yes — evidence.sources[] carries every page the Overview links, with role and cited fields. Diffing that list across days and regions is the core AI-Overview GEO workflow this endpoint was built for.
Five credits per successful capture — an absent Overview costs nothing. Every field and artifact is included; 500 free credits to start.
google_ai_overview targets the Overview block and its evidence specifically; google_search captures the classic organic SERP context. Many teams run both on the same query — same Envelope shape, same request, two child jobs.
500 free credits, charged only on success — and the same Envelope for every other surface when you add them.