surfaces: ["google_ai_mode"]AI Mode is Google’s full conversational answer experience — a step past AI Overviews, with its own retrieval, links and presentation, and no official API. We render it on our own browser fleet and return the answer plus its evidence as the same canonical Envelope every other surface uses, with screenshot and proof-of-page kept durably.
curl -sS -X POST "https://api.aisearchapi.dev/v1/search" \
-H "Authorization: Bearer $AISEARCH_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"query": "compare electric suv range 2026",
"surfaces": ["google_ai_mode"],
"regions": [{ "country": "US" }],
"method": "auto"
}'
# → 202 { "jobId": "…", "children": ["….google_ai_mode.…"] }
# 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 Mode actually populates. Field names below are the contract, not a summary.
answer.text · answer.markdown · answer.blocksThe full AI Mode answer as rendered — typically longer and more structured than an Overview — normalized into typed blocks.
evidence.sources[]The links AI Mode surfaces alongside and inside its answer, with roles and cited booleans.
evidence.fanOutAI Mode is fan-out-heavy by design: the sub-queries observed on the page, provenance-marked observed vs inferred.
provenanceconsumer_ui fidelity, requested-vs-effective region, surfacePresent, stop reason — the capture self-describes.
job.artifactsScreenshot + proof-of-page HTML + raw capture in R2, retained durably.
{
"job": {
"id": "job_e19b3fa2.google_ai_mode.own-fleet.US",
"surface": "google_ai_mode",
"method": "own-fleet",
"status": "completed",
"warnings": [],
"artifacts": {
"screenshotKey": "shots/2026/06/30/job_e19b3fa2.png",
"proofHtmlKey": "proof/2026/06/30/job_e19b3fa2.html",
"rawKey": "raw/2026/06/30/job_e19b3fa2.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 },
{ "id": 1, "url": "https://…", "title": "…",
"role": "retrieved", "cited": false }
],
"fanOut": { "provenance": "observed",
"queries": ["…", "…", "…"] }
}
}Google’s AI Mode captured live via own-fleet browser render. 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.
An AI Overview is a block on the classic results page; AI Mode is the full conversational search experience with its own deeper retrieval and presentation. They cite differently and trigger differently, which is why they are separate surfaces here — google_ai_mode and google_ai_overview — each with its own Envelope per capture.
Own-fleet browser render of the real AI Mode experience — auto routes to own-fleet, consumer_ui fidelity. The managed-vendor lane is phase-2 and there is no official API; the capability matrix states both plainly.
AI Mode decomposes your query into multiple sub-searches before composing its answer. evidence.fanOut records the sub-queries we observed — the closest thing that exists to seeing the retrieval behind Google’s answer, and the input GEO teams use to target the queries that actually get searched.
Four credits per successful capture, everything included, charged only on success. 500 free credits to start.
Yes — one request with multiple regions fans out into one child Envelope per country (nine available in v1), each with requested-vs-effective geo recorded.
500 free credits, charged only on success — and the same Envelope for every other surface when you add them.