Fleak joins the Databricks startup accelerator. See the announcement

Test it on real data. Ship it as a version.

Every Fleak pipeline is tested on sample events before it deploys, published as a versioned DAG, and promoted through your environments like any other code. Roll back in one click. Manage it all from the UI, the API, or Terraform.

Book a demo See how it works
  • Sample testing
  • Versioned releases
  • One-click rollback
  • YAML & REST API
  • Terraform
  • CI/CD

Pipeline changes ship
without a test, a version, or a way back.

Parsers live in config files and SIEM settings. A one-line change goes straight to production, and the first test is the dashboard that stops moving.

No test before production

Most parser edits are validated against the live stream. A typo becomes an outage measured in missing events.

No history

When detections break, nobody can say which change did it or what the pipeline looked like last Tuesday.

No way back

Reverting means rewriting the old config from memory while events keep failing.

Every change tested,
versioned, and reversible.

Every pipeline is a DAG with history. It's tested before deploy, versioned on publish, promoted through environments, and executed deterministically by Zephflow.

Test on real or synthetic samples

Paste events, sample a live topic, or generate synthetic ones. Run one node or the whole DAG and see output and errors per node before anything deploys.

Versioned releases

Publish a version with a description and a commit hash. Only published versions deploy, and every deployment shows which one is running.

Rollback in one click

Switch a pipeline back to a previous version, or clone an old version to start a new edit.

Pipelines as code

Export and import DAGs as YAML, script everything through the REST API and the Python and Java SDKs, and manage pipelines, connections, and destinations with Terraform.

Git and CI/CD

Keep pipelines in Git, review changes in pull requests, and promote them from dev to staging to production through your CI/CD system.

Credentials defined once

Store connection secrets once, encrypted and hidden from everyone but their owner, or pull them from your vault. Scope API keys to exactly what each job needs.

Test before it touches production.

Real samples, per-node errors, versioned releases, and a rollback button.

Read the docs Book a demo

From draft to deployed
in four steps.

01 / Build on samples

Start the DAG from real events: pasted samples, a live topic, a data asset, or synthetic data.

02 / Test node by node

Run any node or the full DAG and inspect the output as JSON or a table, with errors shown per node.

03 / Publish and promote

Snapshot the DAG as a version, review it like code, and promote it from dev to production.

04 / Deploy and watch

Deploy to a cluster, set replicas, and watch throughput and error rate. Roll back if anything moves.

Built like software.
Run like infrastructure.

The Workflow Builder gives every pipeline a test loop, a version history, and a rollback path. Zephflow, the open-source engine underneath, executes the deployed version deterministically, the same way every time.

Manage it from

  • Workflow Builder
  • REST API
  • Terraform
  • Python SDK
  • Java SDK
  • GitHub Actions

A parser change breaks a field on Friday afternoon.
How fast is it back?

Without versions

Hours

Find the change, rebuild the old config from memory, redeploy, and hope nothing else moved.

With Fleak

One click

Roll the pipeline back to the last good version. The fix is tested on samples before it ships again.

Review

Versioned

Every published version has a description and a hash, so the incident review starts from what changed.

Test before deploy, version on publish, roll back in one click. The same discipline your engineers already apply to code.

"A parser is code. It deserves tests, versions, and a rollback button."

Pipeline lifecycle · Fleak Platform

Bring the parser nobody wants to touch.

30-minute working session. We'll rebuild it on samples and show you the test loop.

Read the docs Book a demo

Explore Related

Routing

Route every log by what it's for

SIEM, lake, and archive, each in its own schema.

See Detail →
Log reduction

Less volume, same coverage

Filter, trim, deduplicate, sample, and aggregate in flight.

See Detail →
Replay

Store it cheap, replay it when it matters

A full-fidelity archive, replayed through the same pipeline.

See Detail →