Agent operations

Operate your AI agents.

Kohral is a control plane for autonomous AI agents. Define what each agent does, monitor its conversations, track its cost, and deploy it when you're ready — all from one dashboard.

Build and test agents free.
Pay only when you deploy.

kohral.sh / fleet6 running
Kohral Fleet overview with agents plotted as nodes across runtimes
Runtime
Hosted or self-managed
Credentials
Vaulted, never inline
Deployment
Gated and deliberate
01 — The problem

Teams ship more agents than they can track.

An agent goes live in an afternoon. Six months later nobody knows how many are running, what they cost, or who approved the last change.

Kohral fixes this. Every agent has a status, a cost, and an owner. You see what's running, what it's spending, and who's responsible — all in one place.

02 — The dashboard

One dashboard for every agent

Fleet overview

Every agent visible at a glance, colour-coded by status: running, drafting, or degraded. No more digging through lists to find what's broken.

Live conversations

Read what your agents say as they say it. Step into any conversation, follow tool calls, and intervene when a human is needed.

Telemetry & cost

Messages processed, health signals, and cost per agent across your fleet. Spot runaway spending before the bill arrives.

Credential vault

API keys stored in a vault and referenced — never pasted into a prompt. Preflight checks every key before an agent can launch.

Gated rollouts

Nothing goes live by accident. Deployment stays blocked until intent is confirmed, capabilities are chosen, and every credential checks out.

Guided setup

A three-step interview gives each agent its role, persona, and runtime. Define it once — no need to re-explain in a prompt every week.

Conversations

Watch the work, not just the logs

Every conversation an agent holds is inspectable in real time — the context it was given, the tools it used. When something goes wrong, you see exactly where.

kohral.sh / conversations
Kohral Conversations workspace showing a live agent thread
kohral.sh / metricsupdated 30s ago
Kohral Metrics telemetry dashboard
Metrics

Know the cost of an always-on fleet

Throughput, health, and spend per agent and per period. The data you need to retire a quiet agent or justify a costly one.

03 — How it works

Build first. Deploy when it's ready.

STEP 01

Define

Walk through the guided setup: role, persona, capabilities, runtime. The agent exists as a draft — free, testable, and reviewable.

STEP 02

Wire credentials

Reference vaulted API keys instead of pasting them. Preflight verifies each one and flags anything missing before launch.

STEP 03

Roll out

Confirm the setup and deploy the agent. From that moment it's live, metered, and visible on the dashboard.

04 — Deployment

Your runtime, your rules

Hosted

Recommended

We run the platform and the runtime. Build as many agents as you want at no cost; billing starts the moment one is deployed.

Self-managed

Run agents inside your own infrastructure. Keep every conversation and credential on your side. Kohral manages the dashboard; the runtime is yours.

05 — Pricing

Simple, usage-based pricing

Drafting
Free

Build, interview, and review unlimited agents in hosted mode. Nothing is billed while an agent is still in draft.

Core
€49 /mo

The dashboard — fleet overview, conversations, telemetry, and credential vault for your whole team. Includes your first deployed agent.

Rolled out
from €19 /agent

Metering starts at deployment and stops when you retire the agent. Per-agent cost is visible on the metrics dashboard at all times.

Beta pricing during private beta — talk to us and we'll tailor it to your team.Talk to us