Varqeni Help

What Varqeni does, and what it does not

Updated 11 November 2026

Varqeni is autonomous operations for production systems, starting with data platforms. Its agents watch a change data capture pipeline around the clock, investigate every signal with evidence, and fix only what a person or a written policy allows.

AvailabilityAvailable to everyone with early access to Varqeni.

How it works

  • It watches. Every 30 seconds the watch service reads PostgreSQL, Debezium on Kafka Connect, the schema registry, Kafka and the Kubernetes workloads behind them, through identities that can only read.
  • It investigates. When a fault holds for two checks in a row, Varqeni opens an investigation, collects the evidence, names the cause and rules out the others. A language model reviews the finding; it can disagree, but it never decides.
  • It acts, within limits. On pipelines, two reversible fixes: resuming a paused connector and restarting a task that stayed failed after an outage. On infrastructure, four: restarting a host's metrics exporter, clearing old rotated logs from a full disk, a rolling restart of a hung workload and rolling back a bad rollout. Each runs only after a named approval or under a written policy, on what your administrator lists, and is checked afterwards.
  • It proves every step. Evidence, the cause, any approval and the outcome are sealed into a bundle you can verify later.

What you should see: The console shows the live pipeline, every investigation with its evidence, the actions taken and the policy that allowed them.

What it can and cannot do

Can

  • Watch a change data capture pipeline on self-managed PostgreSQL, find the cause of a fault and show the evidence.
  • Run two pipeline fixes and four infrastructure fixes, after a named approval or under a written policy, and check they worked.

Cannot

  • Drop a replication slot, reset offsets, delete or recreate topics, change a publication or restart a database. Ever.
  • Change infrastructure without a named approval, or run on managed cloud databases yet.

Still stuck? Contact support or email info@fortisdatagroup.com.