What is Causet?
Causet turns fragile workflows into replayable timelines.
Your app submits intents. Causet records events. Queries and projections expose current state. The timeline can be inspected, forked, replayed, or repaired.
Causet sits beside an existing app. It does not replace the frontend, database, queue, vector DB, or services. Adopt it one endpoint, webhook, background job, or agent action at a time.
The problem
Important workflows often live inside API routes and webhook handlers:
validate → write DB → call service → send email → update statusWhen something fails halfway, logs are not enough. Teams cannot reliably answer what completed, what did not, or how to repair state.
AI agents make the same problem sharper: decisions, memory, tool calls, and side effects also need to be inspectable.
AI is not the only use case. It is the sharpest version of the problem.
Causet gives backend workflows and AI agents the same thing: a durable, replayable timeline of meaningful state changes.
What Causet records
request_access
→ access_requested
→ approval_required
→ access_approved
→ notification_sentThat is not a log dump. It is the application timeline that defines how state changed — and what you can replay or repair.
What you define
Developers author .causet files:
| Section | Purpose |
|---|---|
state | Durable workflow entity fields |
events | Immutable domain facts |
actions / intents | Commands from apps, jobs, webhooks, or agents |
projections | Read models derived from events |
queries | Named reads against projections |
sagas / workflows | Multi-step processes |
decisions / memories | Optional AI — same timeline as everything else |
The compiler validates the model. The runtime executes state changes deterministically and commits them to the ledger.
What Causet is not
- Not a replacement for Postgres, Kafka, Redpanda, or your API
- Not a vector database
- Not a generic web framework
- Not only a workflow engine
- Not a logging product
You keep your database, services, queues, APIs, and frontend. Causet is the deterministic state layer for flows where correctness, auditability, replay, and repair matter.
See What Causet handles.
Start here
Not sure if Causet fits? Read When Not to Use Causet.