Platform

    StreamFlow

    The distributed event mesh behind Pulse.

    High-performance event streaming. Kafka compatible. Built for intelligent applications.

    Try it through Pulse See benchmarks
    Why StreamFlow exists

    Building intelligent applications shouldn't require assembling four different distributed systems.

    Most teams assemble a broker, a stream processor, a deployment platform and a state store.

    StreamFlow was designed to remove that operational complexity while delivering the performance expected from modern event infrastructures.

    Built for Pulse

    Pulse is the product. StreamFlow is the engine.

    Pulse
    powered by
    StreamFlow

    Every Pulse app runs on StreamFlow. When you install Pulse, you inherit the mesh — replication, durability, Kafka compatibility, all of it.

    Discover Pulse

    Architecture

    Four layers. One binary. No glue.

    Producers
    Any Kafka client, HTTP, SDK.
    StreamFlow
    Distributed event mesh. Shard-per-core. Raft.
    Agents & Handlers
    Stateful compute. Materialized views.
    Applications
    APIs, dashboards, downstream systems.

    Performance

    Numbers, not adjectives.

    ~9M
    events/sec per node
    Shard-per-core
    no coordination overhead
    Zero-copy
    io_uring hot path
    Kafka :9092
    wire-protocol compatible
    Raft
    deterministic replication
    Already on Kafka?

    Keep your producers.

    StreamFlow speaks the Kafka wire protocol on :9092. Point your existing producers and consumers at StreamFlow and they keep working — no code change, no migration ceremony.

    Migration, in one line.
    // Before
    new KafkaProducer(bootstrap: "kafka:9092")
    ↓
    // After
    new KafkaProducer(bootstrap: "streamflow:9092")
    ✓ Done.
    Drop-in compatibility

    Same engine. Every scale.

    Embedded
    Runs inside your process. Zero infra.
    HA
    Multi-node Raft cluster. Zero downtime.
    Mesh
    Federated regions. Global topics.
    Proof

    Built on StreamFlow

    The engine is credible because it runs products — not because it's fast.

    Available

    Pulse

    Local-first runtime for intelligent event-driven apps.

    Coming

    RevenueOS

    Real-time revenue intelligence for growth teams.

    Coming

    Resonance

    Agentic customer signals across every channel.

    Open

    Your app

    Same engine. Same guarantees. Yours to build.

    Designed for platform teams.

    The people who run the platforms other people build on.

    Platform teams
    Infrastructure engineers
    Distributed systems
    Architects

    Documentation

    Deep dives for platform engineers.

    Architecture

    Shard model, control plane, data plane.

    Wire protocol

    Kafka-compatible frames on :9092.

    Storage

    Segmented log, tiered storage.

    Raft

    Leader election, log replication.

    Replication

    Cross-region, tunable durability.

    Observability

    Metrics, traces, audit stream.

    Benchmarks

    Reproducible scripts, configs, results.

    Don't benchmark the engine. Build on it.

    The fastest way to evaluate StreamFlow is to deploy a Pulse application. Three commands, and you're running on the mesh.

    Start with Pulse Reproducible benchmarks