Frequently asked questions.
Accurate across both ways to run dispatchmy.ai: fully hosted by us, or self-hosted on your own machine.
Do I need my own LLM API key?
Depends how you run it. On hosted, no. LLM credits are included and metered for you (DeepSeek V4 flash and pro today). On self-host, yes, bring your own Anthropic key, OpenRouter key, or any OpenAI-compatible endpoint (Ollama, LM Studio, vLLM, a custom proxy). Either way, you can also add your own key on hosted if you want to.
Can the agent see my API keys?
No. On self-host, your provider and channel keys stay with your daemon and are only used when it proxies a tool call, so even a prompt-injected agent cannot leak a credential it never sees. On hosted, the managed LLM key lives in the container environment, which neither the agent nor you can read, and we meter usage against your credit balance.
What happens to my data?
Self-host: your configs, sessions, and files stay on your machine, we only keep your account details. Hosted: your daemon runs on our infrastructure, so we store the configs and sessions needed to run the service for you. See our privacy policy.
What do I need to install?
Hosted: nothing, it runs in your browser and we spin up the daemon for you. Self-host: Docker on macOS or Linux (amd64 and arm64) as Tier 1, Windows runs via Docker Desktop and WSL2.
Is this built on top of another agent framework?
No. The runtime is our own, built from day one around containerized agents, a configuration dashboard, and a built-in management MCP that lets agents configure and improve themselves. Secret values still stay hidden from the agents.