Apollo

Apollo is an open source “brain” for physical voice agents. It listens when you say its chosen wake name, remembers what you tell it, and takes actions that end in real outputs on your side. You supply the body it speaks through (mic, speaker, and face), and Apollo handles the voice interaction, memory, and task execution.

It’s a fit if you want an agent that works outside a chat window, stays tied to your account, and runs on infrastructure you own. It’s also built for teams and builders who want to connect the agent to their existing tools.

• Voice turns: it answers out loud in short sentences, so you don’t have to type or read.
• Memory: it keeps information you share and brings it back when it’s relevant.
• Reminders: timers and scheduled alerts fire on the physical body.
• Live answers: it can search the internet and return the answer rather than a list of links.
• Coding agent: you describe a change, it works in a sandboxed repo and reports what happened.
• Your tools (via MCP): connect services you already use through MCP.
• Character: the wake name, voice, and multiple speech modes are tied to the face it displays.

Apollo stands out because the “brain” and everything it knows (memory, schedules, character, and the audio context) live in a deployment you own. The body is meant to be swappable, so the agent’s knowledge moves with the brain, not the device. The content also emphasizes that API keys stay in your environment, not in chat, and that there’s no separate account to create or subscription to pay for Apollo itself.