Exclusive: Mecka AI raises $60 million to train robots with human data sourced from body sensors and iPhones - Fortune

2026-06-01 · Source: Series A" OR "Series B" OR "Series C" AI startup via Google News · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Fundamental Awareness, short

Summary

Mecka AI, a New York City-based startup, has secured \$60 million in funding across two previously unannounced rounds: a \$25 million Series A in November and a \$35 million follow-on investment. The company aims to train robots using "embodied" AI data, specifically human physical motions, hand gestures, and walking gaits, collected via custom body sensors and iPhones. Led by Framework Ventures, with participation from Menlo Ventures and others, Mecka AI's cofounders, including CEO Josh Gao, developed an unconventional hypothesis to train robots on human-sourced data rather than traditional teleoperation. The startup, which began operations in 2025 after extensive data collection, projects an annual run rate of \$100 million from signed contracts and plans to actively assist firms in integrating and training robotics models, not merely aggregating data.

Key takeaway

For Directors of AI/ML developing robotics solutions, Mecka AI's approach highlights a critical shift towards human-sourced "embodied" data. You should evaluate integrating diverse human motion data, collected via sensors or mobile devices, into your training pipelines to achieve more generalizable robot intelligence. This method offers a scalable alternative to traditional teleoperation, potentially accelerating deployment of useful robots today. Consider piloting data collection strategies that leverage everyday devices like iPhones.

Key insights

Mecka AI trains robots using human physical motion data collected from body sensors and iPhones.

Principles

• Embodied AI data is crucial for advanced robotics training.
• Human-collected data can scale robotics training beyond teleoperation.

Method

Collect human physical motion data using custom body sensors and iPhones, then use it to train robotics models.

In practice

• Integrate human-sourced motion data into robot training pipelines.
• Explore custom hardware and mobile devices for data capture.

Topics

• Robotics Training
• Embodied AI
• Human Data Collection
• Body Sensors
• iPhone Data Capture
• AI Venture Funding

Best for: Investor, Entrepreneur, Director of AI/ML

Related on AIssential

• Video Friday: Digit Learns to Dance—Virtually Overnight — Advanced robotics are achieving new performance thresholds in control, learning, and human interaction.
• Rivian spin-out Mind Robotics raises $500M for industrial AI-powered robots — Mind Robotics secured $500M to develop AI-powered industrial robots with human-like dexterity for factory automation.
• Korea’s biggest manufacturers back Config, the TSMC of robot data — Asia's manufacturing strength is driving investment into physical AI, exemplified by Config's robotics data platform.
• Physical Intelligence Team Unveils MEM for Robots: A Multi-Scale Memory System Giving Gemma 3-4B VLAs 15-Minute Context for Complex Tasks — MEM provides VLA models with dual-track memory for long-horizon robotic tasks, enabling adaptation and real-time performance.
• The Download: gig workers training humanoids, and better AI benchmarks — AI training and evaluation methods require significant evolution to meet real-world demands and ethical considerations.
• Neura Launches Europe's Largest Physical AI Training Center — High-quality, realistic training data is the primary challenge for advancing intelligent robotics, not hardware.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Series A" OR "Series B" OR "Series C" AI startup via Google News.