Khosla-backed robotics startup Genesis AI has gone full stack, demo shows

2026-05-06 · Source: Robotics News | TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Emerging Technologies & Innovation · Depth: Intermediate, medium

Summary

Genesis AI, a startup that secured a $105 million seed round, has launched its first foundational AI model for robotics, GENE-26.5, alongside a custom robotic hand developed in partnership with Chinese company Wuji Tech. The company adopted a "full stack" approach, realizing the necessity of hardware control for optimal model performance. Their robotic hand mimics human hand size and shape, enabling extensive data collection to train models for complex tasks like cooking, playing piano, and lab work. Genesis AI also developed a sensor-loaded glove for real-world data collection, aiming to bridge the "embodiment gap" by leveraging human data. The model, named GENE-26.5 for May 2026, is expected to undergo rapid iteration thanks to an advanced simulation system. The company, with offices in Paris, California, and London, plans to unveil a full-body general-purpose robot soon.

Key takeaway

For research scientists developing advanced robotic systems, Genesis AI's full-stack approach highlights the critical advantage of integrating hardware design with AI model development. You should consider how custom hardware, like human-mimicking robotic hands and sensor-loaded gloves, can significantly enhance data collection and accelerate model training iterations, ultimately leading to more capable and versatile robotic intelligence.

Key insights

Controlling both AI models and hardware accelerates robotics development by enabling more effective data collection and model training.

Principles

• Mimicking human anatomy reduces the embodiment gap.
• Full-stack control improves model intelligence and iteration speed.

Method

Design human-like robotic hands and sensor-loaded gloves to collect vast amounts of human-based data, then combine with simulation for rapid model iteration and training.

In practice

• Utilize human-mimicking hardware for data capture.
• Employ sensor-loaded gloves for real-world data collection.

Topics

• Genesis AI
• Robotics AI
• GENE-26.5
• Robotic Hands
• Data Collection

Best for: Research Scientist, Robotics Engineer, AI Scientist, Entrepreneur

Related on AIssential

• Can a Prosthetic Hand Deliver Humanoid Dexterity? — Integrating human-proven AI bionic hands into humanoids enhances factory-ready durability and dexterity.
• Video Friday: Heavy Robotic Machinery Operates Itself — Robotics advancements span autonomous heavy machinery, versatile multi-robot systems, and human-robot interaction.
• Cheers to AI: ADAM Robot Bartender Makes Drinks at Vegas Golden Knights Game — Robots trained in simulation with synthetic data can perform complex real-world tasks with high precision and low latency.
• Robots run this laboratory in Japan — and are changing how scientists work — AI-powered laboratory robots can autonomously manage complex biological experiments, accelerating discovery and freeing human researchers.
• Video Friday: AI Gives Robot Hands Human-Like Dexterity — Advanced AI and robotics are enabling human-level manipulation, extreme flight, and complex autonomous inspection.
• Rivian’s RJ Scaringe thinks we’re doing robots all wrong — Mind Robotics aims to redefine industrial automation with AI-powered robots optimized for manufacturing tasks, prioritizing specialized hands over complex human biomechanics.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Robotics News | TechCrunch.