Source Channels
Synorb watches specific source surfaces such as filings, reports, feeds, releases, transcripts, notices, podcasts, and research pages.
A content feed for AI is a structured stream of source-grounded updates that an agent can store, route, cite, and act on. It is most useful when the agent needs repeatable coverage, not a one-off search result.
Synorb provides content feeds through watched Source Channels and Streams. The unit an agent receives is a Manifest: a source-linked object with Briefs, Signals, Records, stable IDs, provenance, and delivery paths for REST, MCP, webhooks, and S3.
Source Channels to Streams to Manifests · REST + MCP + webhooks + S3 · Provenance by default
Search begins after a user asks a question. A content feed starts when a watched source changes. That difference matters for AI systems that monitor companies, markets, policy, research, media, or other current-event surfaces.
Synorb watches specific source surfaces such as filings, reports, feeds, releases, transcripts, notices, podcasts, and research pages.
Streams scope coverage by topic, entity, source set, or saved query so an agent can listen to the areas it is responsible for.
Manifests package the source event, Brief, Signals, Record, source URLs, dates, stable IDs, and delivery metadata.
Synorb is a fit when an agent or application needs recurring, source-grounded context. It is not an arbitrary scrape-any-URL system. If the workflow depends on known sources or durable coverage areas, listening first is usually cleaner than asking an agent to rediscover the web.
| Use Synorb for | Still use web search for |
|---|---|
| Monitoring sources, companies, topics, policies, data releases, filings, research, and current-event Streams. | Unknown sites, exploratory discovery, long-tail questions outside watched coverage, and one-off human browsing tasks. |
| Keeping agent memory current with provenance, stable IDs, and source URLs attached. | Filling gaps when the agent needs a source Synorb has not been asked to watch. |
AI systems can use a Manifest directly in retrieval, ranking, summarization, alerting, review, or workflow automation. The object is designed to carry enough context for both machine use and human audit.
{
"manifest_id": "1777525429698648000",
"stream_names": ["company-monitoring", "ai-infrastructure"],
"source": {
"name": "Observed Source Channel",
"url": "https://source.example/update",
"published_date": "2026-06-17"
},
"brief": {
"title": "Source-grounded update"
},
"signals": [
{
"claim": "Atomic claim extracted from the source event",
"domain": "technology"
}
],
"record": {
"stable_id": "record_1777525429649909800"
}
}
MCP is useful when an agent is exploring Streams, asking for Manifests, or prototyping a workflow. Shipped applications usually call Synorb from server-side code over REST, accept webhooks for new items, or ingest S3 drops for durable storage.
Use Synorb MCP when a compatible agent should inspect Streams, retrieve Manifests, and cite source URLs during development.
Use the docs for API calls and delivery options. Keep credentials on the server side and route feed items into your own storage or workflows.
Streams are priced by plan, quota, and delivery cadence. This page does not repeat the full plan table because pricing changes belong in one canonical place.
A content feed for AI is a structured stream of source-grounded updates that an agent or application can store, route, cite, and use without re-discovering the same source every time.
No. Synorb is built around watched Source Channels, Streams, and Manifests. It helps agents listen to known coverage areas and use web search only when they need to fill gaps.
Synorb returns Manifests: structured objects that can include Briefs, Signals, Records, stable IDs, source URLs, source metadata, Stream routing, and ontology tags.
Yes. Provenance is part of the product shape: source URL, source name, published date, capture date, Source Channel, and lineage travel with the Manifest.
Yes. Use REST, webhooks, or S3 from server-side code for production delivery. MCP is useful for agent-native development and interactive workflows.
Start with keys, read the API docs, or use the build hub if you are connecting Synorb inside an agent-built application.