Source changes
A watched surface updates
A Source Channel emits a new filing, post, report, transcript, notice, release, podcast, feed item, or data change.
Agent data source
Real-Time Content Feeds For AI Agents.
Real-time content feeds keep an AI agent current by delivering source-grounded updates as watched sources change. The agent can listen to known coverage areas, preserve citations, and use search only when it needs open-ended discovery.
Synorb turns watched Source Channels into Streams of Manifests. Agents can call those Streams through MCP, use REST from server code, receive webhooks, or use S3 archive exports for historical backfills and durable workflows.
Listen first · Search for gaps · Keep provenance attached
01 / Freshness
Agents need state, not just answers.
A monitoring agent, research agent, or product workflow should not depend on repeated cold web search for every run. A real-time feed lets the system maintain its own current state from source-grounded events.
Stream routes
Coverage stays scoped
The update is routed into the Stream or Streams an agent, product, or data team already cares about.
Manifest lands
The object is usable
The agent receives a Manifest with Briefs, Signals, Records, dates, source URLs, and stable IDs.
Workflow runs
Action has context
The app can alert, rank, retrieve, summarize, store, or ask a model to reason over the update.
02 / Interfaces
Choose the delivery shape that matches the agent.
Synorb is not limited to one integration mode. Use MCP when the agent is the operator. Use REST and webhooks when your application infrastructure owns live delivery, and S3 archive exports when it needs historical backfills.
MCP
Interactive agents
Connect compatible agents to Synorb MCP so they can list Streams, retrieve Manifests, and cite sources.
REST
Server-owned apps
Call the REST API from backend code when a shipped product needs predictable credentials and request control.
Webhooks / S3
Durable delivery
Push fresh items into workflows or persist drops for batch processing, warehouse ingestion, audit, and replay.
03 / Object
The feed item is a Manifest.
A Manifest gives the agent enough structure to do useful work without losing the source. It can be stored in agent memory, indexed for retrieval, reviewed by an operator, or joined into a warehouse.
Manifest excerptreal-time
{
"manifest_id": "1777525429698648000",
"stream_names": ["policy-monitoring", "company-monitoring"],
"cadence": "live",
"source": {
"name": "Observed Source Channel",
"url": "https://source.example/update",
"published_date": "2026-06-17"
},
"lineage": {
"source_channel_id": "observed-source-channel"
},
"delivery": {
"interfaces": ["REST", "MCP", "webhook", "S3"]
}
}
04 / Limits
Real-time feeds do not replace every search.
Synorb is strongest where the coverage is known or can be defined as Streams. Agents should still search the open web for unknown sources, novel user questions, or background discovery outside watched coverage. The reliable pattern is listen first, search for gaps.
Use Synorb
Known coverage areas
Company monitoring, market updates, current-event feeds, research tracking, source watchlists, and workflows that need citations.
Use search
Unknown destinations
Open-ended discovery, background browsing, exploratory source finding, and questions outside the Streams you have configured.
05 / Build
Start with docs, then ship from the server.
If you are prototyping with a coding agent, use the build hub and MCP setup pages. If you are shipping an app, keep keys off the frontend and call Synorb from trusted backend code.
FAQ
The short version.
What makes a content feed real-time for an AI agent?
The feed is tied to source changes, not only user prompts. Synorb watches Source Channels, routes updates into Streams, and exposes fresh Manifests for agents to retrieve or receive.
How fresh is Synorb data?
Freshness depends on the Stream, source behavior, and delivery cadence. Synorb supports recurring delivery for self-serve plans and Continuous Delivery for Enterprise teams.
How do I connect Synorb to an agent?
Use MCP for agent-native exploration and retrieval. Use REST or webhooks from server-side infrastructure when shipping a product, and use S3 archive exports for historical backfills.
Can Synorb push new items into my application?
Yes. Production workflows can use webhooks for event push, REST for polling or queries, and S3 archive exports for historical backfills and replay.
Does Synorb crawl any URL on demand?
No. Synorb is built around watched Source Channels and Streams. Agents should still use web search for unknown sources or open-ended discovery outside watched coverage.
Test
Test Synorb feeds for free.
Want to connect to Synorb's graph to test source-grounded feeds for free? Start with free test credentials, then connect through Core MCP or REST.
Free test credentialscurl
curl -s https://synorb.com/connect
Start
Connect an agent to a live feed.
Use MCP for interactive agent work, REST and webhooks from your backend for live production delivery, and S3 archive exports for historical backfills.