Building Robots That Feel Alive: Inside Innate’s Mars Launch

Forget backflips. The real frontier in robotics isn’t circus tricks — it’s building machines that feel alive. Earlier this week, Axel Peytavin unveiled MARS, a tabletop companion robot that can already play chess, learn new skills from a community, and even serve as a social lifeline for people who struggle with tech. It’s small (25×25×40 cm), open, and designed for tinkering — a platform where prompting meets robots.

We got an early look inside Axel’s Palo Alto garage while MARS was still a rough prototype called Maurice. What we saw was the birth of an open-source robot kit meant to be teachable, hackable, and a proving ground for embodied AI.

Watch now

Most robot reels right now are about motion, viral backflips, kung fu katas, choreographed dances. Fun to watch, but they don’t scale to real life. The breakthroughs that matter are in manipulation, adaptability, and trust. Folding laundry. Tidying up. Helping a grandparent connect with family.

In the interview, Axel talks about robotics as a platform: not a sealed demo bot, but an open-source kit for tinkerers with an agent framework and training module. His goal is enabling builders to easily experiment with autonomy, embodiment, and long-term skills the kind of groundwork that makes robots companions instead of stunts.

“We’re not impressed by backflips. We want a robot that feels alive — one that learns from its environment and becomes part of your life.”
– Axel Peytavin, Innate Robotics

Innate’s open-source kit

• Wi-Fi + WebSocket control for real-time responsiveness.
• NVIDIA Jetson on-board compute with camera-based learning.
• Simple Python SDK for training and deploying new skills.
• Fully open hardware with custom PCB and modular sensors.

Three frontiers for robotics

• Manipulation over motion: Progress in fine-motor tasks is far more valuable than flashy humanoid gaits.
• Agentic frameworks in the real world: Robots need tool registries, memory, and long-term planning to be trusted.
• Open-source first: Democratizing robotics lowers barriers for builders and accelerates iteration.

Enjoy!