It runs beside your systems, not instead of them.
AnantState reads events and reference data from the systems you already run. It only changes anything in them through named, recorded actions. Your existing systems stay the source of truth.
In plain termsIt sits beside your systems and reads from them. It does not replace them.
Getting events in
| Route | Use when |
|---|---|
| Streaming | You have a live event stream and want entity state to advance in real time |
| REST source | A system can push, or be polled, over HTTP |
| Scheduled pull | A warehouse or batch system holds the history and latency is acceptable |
| File or bulk import | Backfilling a population to stand up a new domain |
Every route lands in the same place: events attributed to a domain, mapped to entities, converted into state updates. Ingest failures are retained in a dead-letter view with the reason, so a mapping bug shows up as a visible backlog rather than as missing data.
What we already speak the language of
A catalog of vetted starting points, not bespoke integrations.
Each preset pre-fills the connection editor with the transport, the auth model, the endpoint and pagination shape, and a suggested entity key. You still supply your instance URL, your credentials and the mapping to your own event schema. Every preset is one of two transports, a REST poll or a SQL source, which is why the list can grow without becoming a maintenance liability.
| Available today | Systems |
|---|---|
| IT service management | ServiceNow, Jira |
| Commerce | Shopify |
| Identity | Okta, and any OIDC-compatible provider |
| HR and workforce | Workday |
| Databases | Postgres-compatible sources |
| Planned | Systems | Why it is not simply a preset |
|---|---|---|
| Cloud data warehouses | Snowflake, BigQuery, Databricks | They need a SQL driver of their own: a real adapter, not a configuration |
| REST vendors with header-based cursors | For example Salesforce | They need a link-following paginator the generic REST source does not yet have |
We publish the distinction rather than blurring it. A marketplace that lists a planned connector beside a working one is how an integration project starts with a surprise. A connector is a convenience, not a prerequisite: if a system can emit an event or answer an HTTP call, it can feed a domain without waiting for us.
How the platform acts: nothing happens without an actor
| Route | Notes |
|---|---|
| Recommended action, human-approved | The default posture. The platform ranks, prices and explains; a person decides. |
| Auto-applied action, where you grant authority | Per action class, within limits and budgets, fully attributed, and halted when evidence conflicts. |
| Notification | Alerting into the channels your teams already use. |
| Export | Entity and alert exports for the systems that still need a file. |
Automation is never a single global switch. See next best action.
A new world is a definition, not a release
A domain definition declares:
- entities and how to identify them
- state dimensions, their bounds, and the primary dimension that carries business value
- baseline dynamics: how state moves without a model
- interventions available, with costs and modeled effects
- the decision horizon the domain must be right at
- the residual weight: how much the learned correction is trusted
Because it is data, a new industry world can be stood up and evaluated without changing the engine, and a definition can be versioned and reviewed like any other governed artifact.
The same engine, in front of many sources
A new world is a definition. The way events arrive changes; the engine does not.
Tell us what you would connect first
Name the system that holds the events. We will say whether it is a preset, a REST source or a bulk import, and what the mapping looks like.