Question 1

What is an autonomous AI agent?

Accepted Answer

An autonomous AI agent is a program that runs on its own, on a schedule, to carry out a goal without you prompting it each time. It reads its instructions, takes actions, remembers what it did, and reports back. With LiberClaw, each agent runs 24/7 on a dedicated VM, wakes on a heartbeat, and behaves according to a plain-markdown SKILL.md file you write.

Question 2

How do I deploy an AI agent that runs 24/7?

Accepted Answer

Create an agent at app.liberclaw.ai, write a SKILL.md describing its behavior, and click deploy. LiberClaw assigns a virtual machine from a pre-warmed pool and brings the agent online with its own HTTPS endpoint, usually in under a minute. From then on it runs on a heartbeat around the clock, with no infrastructure for you to manage.

Question 3

Do I need an OpenAI API key to run an AI agent?

Accepted Answer

No. LiberClaw agents use LibertAI for inference, which runs open models such as Qwen3.6, Qwen3.5 122B, Gemma 4, DeepSeek V4 Flash, Qwen3 Coder, and GLM-4.7. The inference is included, so you do not need an API key from OpenAI, Anthropic, or any other provider.

Question 4

Can AI agents remember things between runs?

Accepted Answer

Yes. Each LiberClaw agent has persistent memory backed by its own database on its own VM. It keeps state across heartbeats, so it can track what changed, avoid repeating work, and build context about your projects over time instead of starting fresh on every run.

Question 5

Is LiberClaw open source?

Accepted Answer

Yes. LiberClaw is fully open source at github.com/Libertai/liberclaw. The agent runtime, the deployment logic, and the platform itself are all public. You can read every line, fork it, or self-host the whole stack.