Start with a task
Frame the workflow around one job such as a visibility snapshot, creator shortlist, or travel demand signal.
Docs
Docs for agent builders
Use ChinaMarketing.ai primitives inside agent workflows, orchestration layers, and machine-paid execution paths.
Updated hourly on supported routes Low-latency request paths 1h to 24h cache defaults
Map the task to primitives
Use one endpoint for quick checks or combine multiple routes when the workflow needs richer context.
Return structured outputs
Feed rankings, snapshots, trend series, and enrichment layers into orchestration and UI decisions.
Registry-ready capability layer
Start with one capability route when the agent needs one complete job contract
$0.63 standard
Brand Visibility Snapshot
/v1/visibility/snapshot
Generate a China brand visibility snapshot with summary assessment, scored strengths and gaps, and priority actions.
Typical wait: about 45s. Live provider-backed visibility runs can take up to 45 seconds. Keep the request open while the system gathers trend, sentiment, and optional weekly snapshot context.
$0.20 standard
Destination Demand Snapshot
/v1/destination-demand/snapshot
Generate a China outbound destination demand snapshot with demand direction, notable signals, risk framing, and priority actions.
Typical wait: about 15s. Destination demand usually returns quickly. Spend context is included when the destination can be mapped cleanly to a supported market.
$0.45 focused
KOL Shortlist
/v1/kol/shortlist
Generate a decision-ready China KOL shortlist with creator candidates, selection rationale, risk flags, and next steps.
Typical wait: about 35s. Shortlist generation can take up to 35 seconds when provider-backed discovery is required. Keep the request open rather than retrying.
Integration patterns
Use the same primitives in different orchestration shapes
Single-call tools
For agents that need one narrow intelligence action with a simple price and response contract.
Multi-call workflows
For systems that stitch together multiple primitives into one branded or task-specific output.
Scheduled monitoring
For repeated checks on a brand, route, creator set, or category over time.
Budget-aware execution
For flows that need predictable per-task cost logic before each agent run.
Recommended starting workflows
Start with a repeatable workflow, then drop into deeper docs as needed
Brand visibility
Combine trends, sentiment, and weekly brand snapshots for a reusable visibility summary.
Travel movement
Use flights and spend signals to build destination monitoring and movement summaries.
Creator discovery
Start with shortlist generation, then add rankings to compare and monitor creator momentum.
Category monitoring
Layer trends, topic rankings, and category snapshots for ongoing market signal checks.
Real response example
China visibility snapshot
A typical agent task can return a compact object that is easy to route into UI, reasoning, or monitoring logic.
Output
{
"insight": "Brand X is showing rising China visibility momentum with medium confidence.",
"key_drivers": [
"Trend momentum is improving on RED.",
"Sentiment is positive across the latest narrative set.",
"Weekly brand snapshot shows above-baseline visibility."
],
"recommendation": "Keep Brand X in the active monitoring set and compare category drift weekly.",
"confidence": "medium",
"source_summary": {
"trend_lane": "qwen",
"evidence_mode": "proxy",
"weekly_snapshot_available": true
},
"cost": {
"amount_usd": "0.63",
"billing_basis": "per_request"
},
"next_action": "Re-run weekly with a category parameter for stronger benchmarking.",
"raw": {
"trends": {},
"sentiment": {},
"brand_snapshot": {}
},
"meta": {
"processing": {
"latency_class": "standard",
"expected_wait_seconds": 45,
"buyer_message": "This route calls live research providers and may take up to 45 seconds. Keep the request open.",
"artifact_retrieval_supported": true
}
}
} API reference bridge
Need schema-level detail, parameter definitions, and endpoint reference?
Use the core developer docs for route-level request and response examples, anchors, pricing, and payment mechanics.
Response design guidance
Use ranking outputs for routing and prioritisation.
Use trend outputs for summarisation and movement checks.
Use snapshots for UI modules, reporting, and planning context.
Use multiple calls when the workflow needs confidence and enrichment.
Use the processing metadata to warn buyers when a provider-backed route may take longer and to point them to artifact retrieval instead of duplicate retries.