Stop building projection infrastructure.
Append events, deploy a WASM function, query materialized state in milliseconds. No Kafka, no consumers, no infra to maintain.
How it works
Four API calls to production.
Create a log
Logs are immutable, ordered sequences of events. Create one per domain concept.
Append events
Send events as raw bytes. They're ordered, never modified, and never deleted.
Deploy a view
Upload a WebAssembly function that processes events into materialized state.
# 1. Create a log
curl -X POST /logs \
-d '{"name": "orders"}'
→ {"log_id": "550e8400..."}
# 2. Append events
curl -X POST /logs/$LOG_ID/append \
-d '{"type":"order.placed","total":99.99}'
→ {"sequence": 1}
# 3. Deploy a view (WASM binary)
curl -X POST /logs/$LOG_ID/views/totals \
-H "Content-Type: application/octet-stream" \
--data-binary @totals.wasm
→ {"view_id": "7c9e2f01..."}
# 4. Query your view
curl -X POST /logs/$LOG_ID/views/totals/query?after=1
→ {"total_orders": 1, "revenue": 99.99}Events are raw bytes. Use JSON, Protobuf, or any format your views understand.
Why Primatomic
Built for engineers who ship
Ship faster
No Kafka maintenance, no Kubernetes clusters, no operational complexity. Just append events and query state.
Data you can trust
Immutable logs with full history. Automatic snapshots. S3-backed archives. Node failures don't lose data.
Performance you can design around
Consistent millisecond reads for materialized views, including tail latency you can depend on.
Use cases
Problems Primatomic solves
Replace your hand-rolled event replay
You're already storing events somewhere. Primatomic gives you immutable logs with built-in replay, so you can stop maintaining your own append-and-rebuild pipeline.
Audit logs that are actually queryable
Append every transaction to an immutable log. Materialized views give you real-time balances and positions without scanning history on every read.
Continuous feature materialization
Stream raw user events into WASM views that maintain up-to-date feature vectors. No batch jobs, no stale data.
Usage metering without the pipeline
Capture product events and materialize billing metrics and usage dashboards. Skip the Kafka-to-warehouse-to-dashboard chain.
Compare
Primatomic vs. alternatives
| Primatomic | DIY (Postgres + app code) | Kafka + consumers | |
|---|---|---|---|
| Time to production | Minutes | Weeks of custom code | Weeks of infra + code |
| Replay from history | Built-in | Build it yourself | Manual offset management |
| Operational burden | None | Your database + app | Kafka, Zookeeper, consumers |
| Consistency model | Read-after-write | Depends on implementation | Eventual |
Reliability
Built for durability
Three layers of persistence ensure your data is always safe
S3 Archives
Every event persists to S3 independently. Replay from any point in history, even after stream retention expires.
Automatic Snapshots
View state checkpoints to S3 at configurable intervals. Fast recovery without replaying entire event history.
Automatic Failover
Cluster coordination ensures exactly-once processing. When a node fails, another picks up seamlessly.
Ready to build?
Start on the free tier. No credit card required. Scale when you need to.