Events and booths
An HEA can answer product questions while Reachy Mini makes the interaction feel alive.
HEA-World for Reachy Mini
Give your AI a physical presence. Give your robot a real identity.
HEA-World connects Human-Enhanced Agents with expressive robots like Reachy Mini, allowing organizations and creators to bring trusted AI personalities into the real world.
Now
Prototype app with mock mode and HEAGuide prefilled for live API tests.
Next
Simulator and hardware testing for safe expressive movement.
Safety
Local adapter owns motion, cooldowns, profiles, and safe stop.
Community
Designed to move into an open-source Hugging Face workflow.
From web agent to embodied channel
Reachy Mini is an open-source expressive robot for AI builders. The HEA-World idea is simple: keep the HEA brain governed and grounded, then let the robot render that same agent with voice, presence, and safe local movements.
One HEA, many renderers
The Reachy Mini integration should not create a second agent or a separate robot brain. A creator builds a normal HEA once. The Reachy channel receives normal answer text and a small semantic cue plan for expression.
😊 warm
🤔 thinking
💡 explaining
✅ confirming
👋 goodbye
Where this becomes useful
The first product goal is not a toy demo. It is a new channel for trusted human knowledge in places where physical presence changes the conversation.
Education
A teacher or institution can make knowledge easier to approach through an embodied guide.
Showrooms
Visitors can ask natural questions about products, services, and next steps.
Community demos
Builders can remix the local cue mapping while HEA-World keeps answer grounding stable.
Architecture in one pass
The robot never receives arbitrary movement instructions from an LLM. HEA-World provides answer text plus semantic cues. The app decides how those cues become safe motion.
1. Visitor asks
Speech or typed text is sent to the Reachy Mini channel session.
2. HEA answers
The existing HEA chat engine and RAG path produce the normal grounded answer.
3. Cue planner enriches
A bounded nano call creates a small emoji-style cue plan for the completed answer.
4. App renders locally
The Reachy app maps cues to safe movement profiles, cooldowns, and stop behavior.
Designed for trust before spectacle
The strongest version of this project keeps the robot expressive while preserving HEA-World's core promise: answers stay grounded in the creator's knowledge, and risky behavior does not leak into the chat engine.
No second brain
Reachy Mini is a channel for an existing HEA, not a separate unmanaged agent.
No raw motor commands
HEA-World emits supported semantic cues; local app code owns motion.
No raw AV storage by default
The API should process conversation text without storing camera or microphone streams.
Prototype roadmap
This is intentionally staged. We can show the idea today, then turn on the real channel API and simulator/hardware validation when ready.
Stage 0
Static app under HEA-World with mock answer and mock robot behavior.
Stage 1
/api/reachymini/chat streams normal HEA answers and cue plans.
Stage 2
Simulator and hardware testing tune safe movement profiles.
Stage 3
Extract the app into a public repo and Hugging Face community workflow.
Want your HEA to become embodied?
Start by building a normal HEA. The Reachy Mini channel is being designed so your existing agent can later speak through a physical robot without changing its knowledge, governance, or core personality.