Figure shows F.04 hands as Adcock argues dexterity needs complexity
The October 3rd preview follows a May design lock for Figure's fourth-generation humanoid, but Figure has not published hand specifications or independent dexterity results.
By Ryan Merket · Published
Primary source: X
Why it matters
Figure's hands are where its general-purpose humanoid ambition meets the engineering tradeoff between dexterity and manufacturability. The preview reveals the bet, while published specifications and repeatable task results will determine whether it works.
Figure founder and CEO Brett Adcock (@adcock_brett) used an October 3rd post to show the hands planned for Figure's F.04 humanoid and make a case for a less tidy tradeoff in robot design: capable manipulation may require accepting more mechanical complexity.
The post offers a look at hardware, not proof of performance. Figure has not published the F.04 hands' degrees of freedom, actuator count, tactile-sensor layout or payload, and there is no independent benchmark showing what they can reliably handle. The central test is whether the design can do varied physical work repeatedly, then be built and maintained at production scale.
Figure is moving from its second-generation F.02 fleet toward F.04. Figure said in its September 30th account of retiring F.02 that taking the robots apart would consume staff time and delay F.04's launch. That puts F.04 on Figure's operating schedule, while leaving its delivery timetable and capabilities unspecified.
A founder's hardware bet
Adcock's willingness to argue for complexity follows an earlier public account of a design decision he regretted. In August, he called Figure's original tendon-based hand an engineering mistake. Figure built and tested that design, then moved on to explore other hand architectures. His account places the hand now on display within a process of making, testing and discarding hardware.
The sequence also reflects Adcock's path as a repeat founder. After co-founding online recruiting marketplace Vettery, he moved into aviation and co-founded Archer Aviation in 2018. A University of Florida alumni profile describes how he sought engineering partners for Archer after Vettery, then helped build a team around a technically demanding hardware business. Figure is his next attempt to turn an ambitious machine into a product people can use.
Figure's F.04 program was already moving before this hands preview. On May 13th, Adcock said the robot had reached full design lock and that parts were starting to ship, according to Humanoids Daily's report. Design lock means the engineering design has reached a defined milestone; it does not establish that production is ready or that the finished robot has demonstrated the promised abilities. The later hand preview adds a piece of the design story without answering those performance questions.
Figure has described its broader goal in its master plan: a general-purpose humanoid capable of human-like manipulation and eventual work alongside people. Its current Figure 03 is presented for household tasks, and Figure's Helix system is intended to connect perception and action in changing environments. A humanoid can walk into a home or factory built for people, but useful work there depends on interacting with objects that vary in shape, position and fragility.
From a preview to repeatable work
A hand's range of motion is only one part of that problem. A mechanically elaborate design can offer more ways to grasp and move objects; each additional mechanism also raises questions about assembly, durability, servicing and control. Figure's post argues for taking on that complexity. It does not show whether the added capability holds up across long runs, varied objects or ordinary wear.
Figure has had real-world hardware experience to draw on. In results published in November 2025, Figure said its F.02 robots spent more than 1,250 hours at BMW's Spartanburg plant, loaded more than 90,000 parts and contributed to production of more than 30,000 X3 vehicles. Those are Figure's reported figures, not an independent evaluation of F.04 or its hands. Figure has also described F.02 as an early deployment whose lessons informed later designs in its BMW account.
Figure announced more than $1 billion in Series C commitments in September 2025 at a $39 billion post-money valuation, led by Parkway Venture Capital. The named participants included Brookfield, NVIDIA, Macquarie Capital, Intel Capital, Align Ventures, Tamarack Global, LG Technology Ventures, Salesforce, T-Mobile Ventures and Qualcomm Ventures. Figure described the round as commitments; that figure should not be read as confirmation that every dollar had closed. Capital can fund repeated hardware iterations and manufacturing capacity. It cannot substitute for evidence that a complex hand performs reliably outside a demonstration. Figure's Series C announcement lists the round's terms and investors.
RuntimeWire has reported on Figure's push toward scale, including its reported 1,000th robot milestone and its Nscale compute agreement. The F.04 hand preview points to a constraint in that buildout: the robot's usefulness depends on details that are difficult to validate with a photograph.
For Adcock, the practical test is whether the mechanical complexity he now favors produces a more useful robot without making it too difficult to manufacture, service or operate. Figure's October 3rd post makes the design bet visible. The specifications and repeated task results that would settle it remain to be shown.