Try Hanzo
Platform

One platform. Every layer.

Models, agents, the record they leave, and the machines underneath. One API for all of it.

The system

One OS. Every layer of the company.

From the company down to the machines under it, with learning and governance across every layer.

One system from the first prompt to production.

Ask a question, ship a change, launch a product, or run the stack yourself. Every path opens onto the same models, agents, data, and policy.

Agent runtime

Agents with hands.

Enso decides. Agents coordinate. Tabs gives them somewhere secure to act.

tool     post_v1_exec_run
subsystemexec · runs in a sandbox
round    2 of 8 · the eighth gets no tools
actor    acme/z · billed here
arguments{"lang":"py","code":"…"} · redacted
result   returned to the modelok
bounds · 30s a call · 90s a run · 3 agents deep

one tool dispatch, as the run’s span records it · sample values

Autonomous within boundaries. Accountable everywhere.

Every action leaves the same record, whoever or whatever started it.

actor   acme · service principal
auth    api-key
action  automations.tool.call
resourceautomations/deploy
seq     #4821 · append-only chainverified

one audit record · sample values

Every tool has a permission, every secret a policy, every execution a boundary, and every consequential action an approval when you require one. Identity, authorization, secrets, isolation, networking and audit are built into Hanzo OS.

Read the security architecture

See the whole company think.

One trace connects the objective that started the work to everything that followed.

Modelsagentstoolsinfrastructureresultcostevaluation

Logs, metrics, traces, sessions and scores share the same operational context.

Explore Observability

Everything works better together.

Thinkwith Enso.
Buildwith App and Dev.
Workin Team.
Learnwith Insights.
Runon Hanzo Cloud.
$curl -fsSL https://hanzo.sh | sh
$hanzo auth login
$hanzo # a coding session in your repo

install, sign in, start working

"The wise engineer owns their tools, lest their tools own them."

Principle 14

Where it runs is your call

Same platform, same console, same API. You pick the machines.

Self-Hosted

One command on a server you already pay for. The whole platform, on hardware you can walk up to.

  • Runs anywhere Docker runs
  • Scales out across a Swarm cluster
  • Drives a Kubernetes cluster you own
View on GitHub

Hanzo Cloud

The same platform, run for you. Sign in and deploy.

  • The machines are already running
  • Deploy on push, or only on a tag
  • Metered and billed per organization
Explore Cloud

Hybrid

Run the control plane on your own server. It deploys to the servers and clusters you already have.

  • Servers stay where they are
  • One console over all of them
  • Per-app target: this host, Swarm, or Kubernetes
Contact Sales

Open source, all of it

Apache-2.0. Forked from Dokploy, and it says so in the NOTICE.

Apache-2.0

Use it commercially. Change it. The upstream attribution travels with it.

It is your server

Builds, images, databases and backups sit on machines you chose, in a region you picked, under credentials you hold.

Read the code

The code behind every deploy is right there, and so is the OpenAPI document the console itself calls.

Open Source Revenue Sharing

Up to 5% of compute goes back to open source

Every deployment is SBOM-verified. Contributors to Dokploy earn a share of compute revenue — transparent, on-chain, and customizable by the community.

Start building

The whole platform is on GitHub. Clone it and run it.

Open source

License: Apache-2.0hanzoai

Get Platform

PaaS — deploy from git