Fleak joins the Databricks startup accelerator. See the announcement

Collect and filter at the source. Manage the fleet from one place.

Fleak's edge agent runs on servers, endpoints, containers, and plant-floor gateways. It discovers what to collect, drops noise before it crosses the network, and normalizes on the spot, all configured and upgraded centrally.

  • Windows
  • Linux
  • macOS
  • Kubernetes
  • OT gateways
  • Fleet management

Five agents per host,
and nobody owns all of them.

A SIEM forwarder, a metrics agent, a log shipper, and a vendor agent each collect overlapping data, each with its own config and upgrade cycle.

Agent sprawl

Every tool ships its own collector. Hosts run several, all reading the same files, all competing for CPU.

Noise crosses the network

Everything is shipped raw to a central tier before anything is filtered, so you pay for bandwidth and ingest you never use.

Upgrades take a quarter

Rolling a config change or a new version across thousands of hosts is a change-management project every time.

One agent,
every host, one console.

The edge agent runs the same deterministic engine as the pipeline, so filtering, parsing, and normalization can happen wherever it is cheapest.

Runs everywhere

Windows, Linux, macOS, Docker, and Kubernetes through a Helm chart, plus industrial gateways on the plant floor. Small enough for the hosts you already have.

Auto-discovery

The agent finds log files, event channels, containers, and metrics on each host and proposes what to collect.

Process at the source

Filter, sample, mask, and normalize before data leaves the host. What you drop at the edge never crosses the network or hits a meter.

Central fleet management

Group agents, push versioned configs, and roll out upgrades across the fleet in minutes, with health and status for every node.

Replace legacy forwarders

Retire the Splunk Universal Forwarder, Beats, Fluent Bit, and vendor shippers. One agent sends to every destination in the format each expects.

Built for restricted networks

Agents connect out over a single encrypted connection, so segmented, regulated, and air-gapped sites need no inbound firewall holes.

Agentless where agents can't go

Collect from servers, appliances, and OT devices without installing anything, through agentless collectors that sit next to them.

Buffer, never drop

Agents and edge nodes buffer locally when the network or a destination goes down, then deliver in order. Nodes mesh with load balancing and failover.

Signals at the edge

Flag anomalies, silent devices, and known-bad indicators as data is collected, and encrypt or mask sensitive fields before anything leaves the site.

One agent instead of five.

Collect once at the source, filter before the network, and send everywhere from the same agent.

The Brain configures.
The agent executes at the edge.

The Brain proposes what each host should collect and how to shape it. The agent runs the approved config deterministically on the host, so the same filtering and normalization logic runs at the edge and in the pipeline.

Replaces

  • Splunk Universal Forwarder
  • Elastic Beats
  • Fluent Bit
  • Fluentd
  • Vendor log shippers

A Windows fleet sends every Security event to the SIEM.
What changes with an edge agent?

Before

Ship it all

Every event crosses the WAN raw, gets parsed centrally, and most of it is never read by a detection.

At the edge

Filtered

Noisy event IDs are dropped on the host, the rest is normalized to OCSF, and only signal leaves the building.

Operations

One console

Config changes and upgrades roll out from one console, versioned, with rollback if a group misbehaves.

Process data where it's produced. Ship only what each destination needs, from one agent you manage in one place.

"The cheapest byte to process is the one that never leaves the host. Filter at the edge, normalize at the edge, ship only signal."

Edge collection · Fleak Platform

Bring your forwarder inventory.

30-minute working session. We'll show what one agent replaces and what it stops shipping.

Explore Related

OCSF

Every source in OCSF, long tail included

211 templates, plus AI-generated mappings for the rest.

See Detail →
Enrichment

Context in every event

Identity, asset, threat intel, and geo context in flight.

See Detail →
PII masking

Mask sensitive data in flight

Detectors, custom patterns, hashing, and quarantine.

See Detail →