AIToday

Hezo launches a self-hosted platform that runs teams of AI agents on your own machine while keeping your API keys and secrets completely hidden from the agents themselves.

Hacker News22h ago3 min read

Key takeaway

Hezo is a self-hosted platform that orchestrates teams of AI agents to complete work while keeping your API keys and secrets completely hidden from the agents. Instead of sharing real credentials, agents see only placeholders, and the platform swaps in the genuine key only when making approved network requests. This lets businesses automate multi-agent workflows without exposing secrets to the cloud or to the agents themselves.

Summaries like this, in your inbox every morning.

Sign up free →

3 Key Points

  • What happened

    Hezo is a self-hosted system where AI agents work in isolated containers on your hardware, never seeing your real secrets—API keys are swapped in only at the network edge when the agent reaches an allowed host, then immediately discarded.

  • Why it matters

    Businesses concerned about exposing credentials to AI systems now have a way to delegate work to agent teams without handing over keys or sending data to a vendor's cloud; you control the machine, the spend, and the audit trail entirely.

  • What to watch

    Hezo works with any major model provider (Claude, ChatGPT, Gemini, DeepSeek, and others via OpenRouter), supports hard budget caps per agent and project, and is installed via a single command at localhost:3100.

FAQ

Can the AI agents see my API keys and secrets?
No. Agents only see placeholders in their container; the real key is substituted only at the network edge when the agent reaches a host you have explicitly allowed, and never appears in the agent's context.
Does my data get sent to Hezo's cloud or to a vendor?
No. Hezo is self-hosted on your machine, so your data stays in your instance; agents reach only the model providers and hosts you have allowed.
How do I prevent agents from running up a huge bill?
You set daily, weekly, or monthly budget caps per agent and per project; agents pause when a budget window is exhausted and resume when it rolls over.

Discussion

No comments yet. Be the first to share your thoughts!

Log in to join the discussion

Related Articles

Stay ahead with AI news

Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.

Get Started Free

Free · takes 30 seconds · unsubscribe anytime

5 minutes a day. The AI essentials.

200+ sources · Email / LINE / Slack

Get it free →