Figure Founder Calls Tendon-Driven Robot Hand His Worst Engineering Decision

On August 13, Figure founder Brett Adcock posted a retrospective on X admitting that the first-generation tendon-driven hand designed in 2022 for the F.01 humanoid was his worst engineering decision in four years. The biomimetic approach theoretically allowed the forearm to house more actuators and achieve more degrees of freedom, but testing led him to characterize it as a local optimum that limited the hand's real-world performance. The public admission spread quickly through the robotics community and ignited debate over dexterous-hand drive architectures.

Confirmed

• Brett Adcock stated in his original post that the tendon-driven design was his worst engineering decision in four years
• The original rationale: the biomimetic design offered more space for actuator placement and more degrees of freedom, and felt intuitively natural
• Per @CyberRobooo, the hand was built and tested in early 2023 and the team once regarded it as an engineering artwork, though it later proved limited
• Adcock characterized the tendon approach as a local optimum

Community reaction

• Per @MarwaEldiwiny and @CSProfKGD, the post sparked debate among roboticists over hand design, covering trade-offs between direct drive and tendon drive, with some arguing that multiple approaches should coexist
• @chrisjpaxton noted that Boston Dynamics' Atlas uses a three-finger design with large brushless motors placed directly in the joints, contrasting with the tendon route
• @CyberRobooo also noted that global shipments of dexterous hands remain only in the tens of thousands

Unconfirmed

• Some reposts described Adcock as a former engineer or former employee, inconsistent with the founder title used in his own post and most reposts; his current role should follow the primary source

Why it matters

• Dexterous hands are a critical component for humanoid robot commercialization, and a top company founder's public post-mortem of a route misstep offers a rare first-hand lesson for the industry
• No consensus exists between direct-drive and tendon-driven routes; Figure's reflection may influence other vendors' technology choices

2026-08-13 ~ 2026-08-15 · 5 related posts

• [source] Figure's Founder Admits Tendon-Driven Hands Were a Major Engineering Mistake — adcock_brett · 2026-08-13
• Former Figure engineer calls tendon-based hand his biggest mistake; Atlas uses direct-drive joints — chris_j_paxton · 2026-08-14
• Figure founder calls tendon hand his biggest mistake; dexterous hand shipments stay tiny — CyberRobooo · 2026-08-14
• Figure ex-engineer calls tendon hand design his biggest mistake, sparking robotics debate — MarwaEldiwiny · 2026-08-15

1 near-duplicate retellings: CSProfKGD