Kobal/Act
Act

Move from insight to action without losing control.

Kobal’s agents gather evidence, determine impact, prepare the response, and route the work. You decide which actions stay off, require approval, or run automatically.

See how it works

A team of agents works the problem before you ask.

When Kobal finds something important, it does not stop at the signal. Specialized agents gather the context, test the evidence, work out the impact, and prepare the response.

Investigation · running now

Refund requests up 4× on the Pro plan.

Investigating
01

Alert Agent

Detected billing complaints at 4× their usual baseline.

02

Clustering Agent

Found failed Pro-plan renewals across 47 tickets.

03

Risk Radar

Identified two affected accounts approaching renewal.

04

Root Cause

Checked releases, known issues, and linked engineering work.

05

Business Impact

Attached account value and €8.4k in affected MRR.

06

Decision Assistant

Recommended the response and its priority.

07

Drafting Agent

Prepared the Slack handoff and account-owner message.

08

Workflow Execution

Routed the approved work and scheduled monitoring.

Prepared responseEngineering handoff, account-owner notice, and a seven-day monitor.
Review actions

15 specialized agents. One support operation.

The agents stay out of the way during daily work. Their findings, drafts, and decisions appear wherever you need them.

UnderstandPrioritizeActImprove operationsBusiness impact
Command Center

See what every agent is doing. Control what it may do next.

Inspect every run, its evidence, quality, cost, and result. Autonomy is configured per action, so safe internal work can advance without giving the system permission to speak for your company.

Agent controlALL SYSTEMS HEALTHY
15agents
6active now
94%quality
7approvals
ActionModeTrack record
Tag and link ticketsAuto206 of 211 approved
Route internal handoffsSuggest41 of 47 approved
Customer-facing messagesOffAlways requires a person

Kobal doesn’t stop at the alert. It follows the outcome.

Every finding can become an investigation, an approved action, and a measured result. Kobal reruns the original check and tells you whether the problem improved, changed, or needs attention again.

  • Every finding carries its evidence and the account names behind it.
  • Every action records who approved it, where it was written, and what came of it.
  • Every fix is measured against the issue it was meant to solve.
ResolvedAlert Agent · monitored 7d
Billing spike, down 71%.
Investigation closed. Result written back to the two accounts at renewal.
Fix approved · Jul 26Volume normal · Aug 2

You approve everything. Kobal earns the rest.

Autonomy is set per action, never switched on by default. Kobal starts by telling you, and only acts on its own once you have seen it get things right.

Level 1 · Off

It watches and tells you.

Detection only. Nothing leaves the app without a person.

Level 2 · Suggest

It drafts, you approve.

Every action arrives prepared with its evidence and a preview.

Level 3 · Auto

It runs the safe ones itself.

Let proven, low-risk work run automatically, action by action.

Every dismissal is data. Your reason feeds the agent’s quality record and its permission to act.