Persona AI
Last reviewed
Jun 9, 2026
Sources
20 citations
Review status
Source-backed
Revision
v2 · 3,903 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 9, 2026
Sources
20 citations
Review status
Source-backed
Revision
v2 · 3,903 words
Add missing citations, update stale details, or suggest a clearer explanation.
| Persona AI | |
|---|---|
| Persona AI, Inc. | |
| Type | Private |
| Industry | Robotics, Humanoid robots, Embodied AI |
| Founded | May 2024 |
| Founders | Nic Radford, Jerry Pratt, Jide Akinyode |
| Headquarters | Houston, Texas, United States |
| Other offices | Pensacola, Florida |
| Key people | Nicolaus Radford (Co-Founder, CEO), Dr. Jerry Pratt (Co-Founder, CTO), Jide Akinyode (Co-Founder, COO), Dr. Matt Carney (Chief Engineer) |
| Industry focus | Shipbuilding, manufacturing, construction, oil and gas, energy infrastructure |
| Total funding | $27 million pre-seed (May 2025), additional $3 million from POSCO Group (December 2025) |
| Business model | Robotics-as-a-Service (RaaS) |
| Website | persona.ai |
Persona AI is an American robotics company headquartered in Houston, Texas, that designs industrial humanoid robots for physically demanding work in shipyards, factories, construction sites, and energy infrastructure. The company was founded in May 2024 by Nic Radford, Jerry Pratt, and Jide Akinyode, three engineers whose careers span NASA's humanoid programs, the DARPA Robotics Challenge, and the first generation of commercial humanoid startups.[1][2]
The company raised a $27 million oversubscribed pre-seed round on May 14, 2025, co-led by Unity Growth and Tides Ventures, with participation from a long list of strategic and financial backers including Mirae Asset Group, Embedded Ventures, Goose Capital, and K2 Global.[3][4] On the same week, Persona AI announced a development agreement with HD Hyundai, the world's largest shipbuilder, to deploy humanoid welding robots in Korean shipyards beginning in 2027.[5] Later in 2025, POSCO Group, the South Korean steel and engineering conglomerate, added a further $3 million investment aimed at piloting Persona AI's robots in steel mills and heavy-industry sites.[6]
Unlike many of its better-known peers, including Apptronik Apollo, Figure AI, and Boston Dynamics, Persona AI has chosen a tighter early market focus. Rather than chasing warehouse logistics or eventual home-robotics use cases, the company is positioning its platform as a rugged, modular industrial worker capable of welding, grinding, painting, and inspection in environments where heat, dust, sparks, and confined spaces routinely defeat conventional robotic arms.[2][7]
Persona AI brings together three robotics careers that converge on the same hard problem: getting a humanoid robot to do useful, repetitive, sometimes dangerous work in the real industrial economy. Each founder has spent the better part of two decades on humanoids long before they became fashionable, and the company's pitch leans heavily on that combined depth.
Dr. Jerry Pratt is the company's chief technology officer. He spent twenty years at the Florida Institute for Human and Machine Cognition (IHMC) in Pensacola, where he led research on bipedal locomotion and dynamic balance that became foundational reading in the humanoid field. In 2015, his IHMC team finished second out of 23 teams at the DARPA Robotics Challenge Finals using a Boston Dynamics Atlas robot, and first among teams whose robots primarily walked on two legs. He earned a B.S., M.Eng., and Ph.D. from MIT, with his doctoral work focusing on passive dynamic walking.[8][9]
In 2017 Pratt co-founded Boardwalk Robotics, a humanoid spinout from IHMC. In 2022 he left Boardwalk to become CTO of Figure AI, the California humanoid startup that has since become one of the most heavily funded companies in the category. He departed Figure AI in May 2024, citing in interviews the geographic strain of commuting between Pensacola and the Bay Area every two weeks, and almost immediately announced Persona AI alongside Nic Radford.[10][11] He is a member of the Florida Inventors Hall of Fame.[8]
Nicolaus "Nic" Radford is the company's chief executive officer. He spent roughly fourteen years at NASA's Johnson Space Center in Houston, working in the Dexterous Robotics Laboratory. There he served as deputy and chief engineer on Robonaut 2, the humanoid torso that flew to the International Space Station in 2011, and as principal investigator for NASA Valkyrie (also called R5), NASA's full-body humanoid built for the 2013 DARPA Robotics Challenge and intended as a precursor to Mars surface robotics.[12]
In 2014 Radford left NASA to co-found Houston Mechatronics, which later rebranded as Nauticus Robotics. Nauticus built the Aquanaut, an autonomous shape-shifting underwater vehicle capable of operating at depths beyond 3,000 meters, and went public on Nasdaq via a SPAC merger in 2022 under the ticker KITT. Radford served as CEO until he stepped down in 2023. He also founded Jacobi Motors, a company spun out to commercialize variable-flux electric motor research from his graduate work at Purdue University.[12][13]
Jide Akinyode is Persona AI's chief operating officer. He worked alongside Radford at Houston Mechatronics and Nauticus Robotics, where he led engineering operations on the Aquanaut program, and brings a background in maritime robotics and program management to a company that is intentionally aiming first at maritime customers.[2][4]
Dr. Matt Carney serves as chief engineer. Carney was previously a research scientist at MIT, where he worked on humanoid actuators and on the design of prosthetic limbs with the Biomechatronics group, and brings actuator design and mechatronics depth to the founding team alongside the three co-founders.[14]
Most of the well-publicized humanoid companies in 2024 and 2025 chose general logistics or eventual home use as their primary go-to-market story. Persona AI explicitly did not. From its first public statements, the company framed itself around the heaviest, dirtiest, and most labor-constrained corners of industry, where skilled workers are scarce, training is long, and injury rates are high.[2][7]
In an interview with The Robot Report shortly after the pre-seed announcement, Radford likened Persona AI's design philosophy to a Ford F-150 pickup truck, contrasted with what he characterized as less rugged demonstration robots elsewhere in the industry. The framing was deliberate: a humanoid that can survive a working shipyard or steel mill has to deal with welding spatter, abrasive dust, ambient heat, water, and the occasional dropped tool, and Persona AI argued that its competitors had been optimizing for clean warehouse and demo-room conditions instead.[1]
Persona AI's public materials list four primary industrial verticals.
| Vertical | Representative tasks | Why a humanoid form factor |
|---|---|---|
| Shipbuilding | Welding curved hull plates, grinding, painting, work inside confined ballast tanks and double bottoms | Workspaces are designed for human bodies and human tools; geometries vary block by block, making fixed automation impractical |
| Heavy manufacturing and metals | Machine tending, billet handling, inspection inside steel mills and forging shops | High ambient temperatures and irregular layouts that have historically been difficult to automate with conventional gantries |
| Construction | Modular building assembly, fastening, lifting, repetitive site work | Construction sites change daily; humanoid mobility allows the same robot to follow the work as it moves |
| Oil, gas, and energy infrastructure | Inspection, valve operation, equipment servicing in refineries and offshore platforms | Hazardous environments where removing humans from confined or high-temperature spaces has direct safety value |
The through-line is that all four sectors are facing severe and well-documented shortages of skilled tradespeople. U.S. shipyards alone need tens of thousands of welders to meet committed naval and commercial orders over the rest of the decade, and South Korean shipbuilders have been openly recruiting from Southeast Asia and India to fill welding lines. Persona AI's wager is that a single hardware platform, tuned by task-specific tooling and software, can take meaningful chunks of that work without requiring customers to redesign their facilities.[5][6]
Persona AI has been deliberately quiet about full specifications. The company has not published a numbered product release, and as of the most recent public disclosures it was still in the prototype phase, with first deliverable humanoids targeted for end of 2026 under the HD Hyundai program. The information that is public is summarized below; anything not in the table has not been verified by the company.[1][5][7]
| Specification | Disclosed value |
|---|---|
| Form factor | Full-body bipedal humanoid |
| Approximate height | 5 ft 8 in (1.73 m) |
| Hands | Dexterous end-effectors compatible with standard industrial tools (welding torches, grinders, hand tools) |
| Platform philosophy | Modular hardware designed to swap tool packages and accommodate task-specific payloads |
| Operating envelope | Designed for indoor and outdoor industrial use, including hot, dusty, and confined environments |
| Control modes | Autonomous task execution plus teleoperation for supervisory and data-collection roles |
| First prototype delivery target | End of 2026 (HD Hyundai program) |
| Commercial deployment target | 2027 onward |
The company has not publicly disclosed payload capacity, degree-of-freedom count, battery runtime, top walking speed, or onboard compute. Renderings on the company website and on partner press releases show a generally human-proportioned bipedal platform with rounded shoulders, two arms, and articulated hands, but published photography of a working prototype remains limited.
Where Persona AI has talked specifications, the emphasis has been on toughness rather than on raw athletic capability. The company has repeatedly described the design target as something able to take a beating in a shipyard or steel mill, work an entire shift, and survive incidental contact with welding sparks, hot surfaces, and abrasive media. That framing implicitly distinguishes Persona AI from companies whose public demonstrations have leaned on parkour, sprinting, or back-flips.[1][7]
Persona AI describes its software approach as a combination of learned policies, model-based control, and teleoperation. The company has not released detailed papers on its stack, but the public picture so far includes the following elements.[1][7]
The company's robots are designed to operate under multiple control modes on the same hardware. For tasks that are tightly defined and repeated, such as welding a specific seam geometry on a hull block, the robot is intended to run autonomous policies trained from a combination of simulation, human demonstration, and on-site fine-tuning. For tasks that change frequently or require human judgement, the robot can be teleoperated by a remote worker, with the robot still handling balance and low-level control while the human handles the perception and decision loop.
Under the HD Hyundai agreement, Persona AI is responsible for the humanoid hardware and for the AI control and learning algorithms, while HD Hyundai Robotics is providing welding-path AI training data drawn from years of shipyard automation, and HD Korea Shipbuilding and Offshore Engineering (HD KSOE) is providing field engineering data and access to live shipyard environments for validation. Vazil Company, a Korean robotics integrator, is responsible for the welding tools and for the industrial test environment.[5] The data partnership matters because welding paths in shipbuilding are highly conditional on plate geometry, joint preparation, gap width, and position, and a large library of professionally welded reference paths is more valuable than any single model architecture.
Matt Carney and Jerry Pratt's combined background in actuators, dynamics, and biomechatronics points toward an integrated approach in which the same controllers that keep the robot balanced also expose contact-rich behaviors to the higher-level planner. That sort of integration has been a recurring theme in Pratt's published research at IHMC on momentum-based whole-body control and capture point planning.[8]
The partnership with HD Hyundai is the central commercial story for Persona AI in its first year. South Korea is the world's largest shipbuilding economy, and HD Hyundai's affiliates collectively build more large vessels than any other group in the world. Welding is the dominant labor input in shipbuilding, accounting for a large fraction of total man-hours on a typical commercial hull or LNG carrier, and skilled welders are increasingly difficult to hire across the Korean shipbuilding cluster.
The formal memorandum of understanding was signed on May 7, 2025, in Houston, Texas, only a week before Persona AI announced its $27 million pre-seed round. The signing parties were HD Korea Shipbuilding and Offshore Engineering (HD KSOE), HD Hyundai Robotics, Persona AI, and Vazil Company, a Korean welding and robotics integrator.[5]
| Party | Role |
|---|---|
| Persona AI | Develops the humanoid hardware and AI-based control and learning algorithms |
| Vazil Company | Builds welding tooling and constructs the industrial test environment for the program |
| HD Hyundai Robotics | Contributes welding-path AI training data and performance validation expertise |
| HD KSOE | Provides field engineering data and access to operational shipyard environments |
| Milestone | Target |
|---|---|
| MoU signing | May 7, 2025 |
| Prototype humanoid welder delivery | End of 2026 |
| Field testing in HD KSOE shipyards | 2027 |
| Commercial deployment | 2027 onward |
In statements accompanying the signing, HD KSOE executive Dong-ju Lee said that the program is intended both to boost productivity and to reduce strain on workers, while Radford described the shipyard as one of the largest real-world proving grounds available for industrial humanoids. Young-hoon Song, vice president at HD Hyundai Robotics, emphasized that the robots in question must observe, reason, and make decisions rather than simply replay welding paths.[5]
The HD Hyundai relationship matters in part because it sidesteps the validation problem that has dogged every humanoid startup: getting a prospective customer to commit to a real production environment rather than a fenced-off pilot cell. By starting with a named program tied to a global shipbuilder, with explicit roles for tooling, training data, and field engineering, Persona AI gave investors a more concrete unit-economics story than is typical for a pre-seed humanoid pitch.
Persona AI's funding to date is concentrated in two events: the May 2025 pre-seed round and the December 2025 strategic investment from POSCO Group. Earlier reporting also referenced a smaller seed-style raise in 2024 that was later expanded as the company built out its founding team.[15]
| Date | Round | Amount | Lead investors | Notes |
|---|---|---|---|---|
| 2024 | Initial pre-seed | ~$10 million (reported) | Undisclosed | Houston-area angels and early strategic backers; reported in InnovationMap[15] |
| May 14, 2025 | Pre-seed (oversubscribed) | $27 million | Unity Growth, Tides Ventures | Round was upsized from initial target; closed within days of HD Hyundai signing[3][4] |
| December 2025 | Strategic | $3 million | POSCO DX ( | Tied to joint development for steel mills and heavy industry[6] |
The full list of named investors in the $27 million pre-seed, drawn from the official PR Newswire release, is unusually long for a round of this size, reflecting both the oversubscription and the company's deliberate breadth of strategic ties:[3]
Mirae Asset is one of the largest financial groups in South Korea, and its participation alongside the HD Hyundai program was widely read as a Korean-side bet on Persona AI as a strategic supplier into the Korean industrial base.[3][4]
In December 2025, South Korea's POSCO Group invested $3 million in Persona AI as part of a broader push by POSCO DX, the group's digital and AI subsidiary, to apply physical AI in steel mills, logistics yards, and other heavy industrial sites. POSCO DX led $2 million of the investment through its corporate venture fund, with an additional $1 million from a POSCO Capital scale-up fund. The agreement included a joint development track aimed at piloting Persona AI's robots in POSCO facilities for tasks such as inspection and material handling.[6]
Persona AI has publicly described its commercial model as Robotics-as-a-Service (RaaS), with customers paying a recurring fee for robot uptime and supported tasks rather than buying a humanoid outright. The RaaS approach lowers the upfront capital that a shipyard, mill, or contractor has to commit, and it gives Persona AI a stronger position on data and software updates over the life of the deployment. The same model has been adopted, in different forms, by Agility Robotics for its Digit warehouse robot and by some service-robotics companies.[2]
The humanoid robotics market in 2025 had at least a dozen serious entrants in North America alone, plus a growing field of Chinese competitors. Persona AI's position is distinctive less because its hardware is radically different and more because of its target customer profile and its founding team's prior experience.
| Company | Robot | Primary target market | Headquarters | Funding scale (latest public) | Notable founders or roots |
|---|---|---|---|---|---|
| Persona AI | Industrial humanoid (unnamed) | Shipyards, steel, construction, oil and gas | Houston, Texas | ~$27M pre-seed (May 2025) plus $3M POSCO (Dec 2025) | NASA Valkyrie, IHMC, Figure AI, Nauticus Robotics |
| Apptronik | Apollo | Logistics, automotive manufacturing | Austin, Texas | ~$1B total (Series A + extensions through Feb 2026) | UT Austin HCRL, NASA Valkyrie SBIRs |
| Figure AI | Figure 02, Figure 03 | Logistics and, increasingly, home | Sunnyvale, California | $1.5B+ total, ~$39.5B valuation reported (late 2025) | Founder Brett Adcock; Jerry Pratt was prior CTO |
| Boston Dynamics | Atlas (all-electric) | Automotive (Hyundai), research | Waltham, Massachusetts | Hyundai subsidiary (private division) | MIT Leg Lab heritage |
| Agility Robotics | Digit | Warehousing and logistics | Salem, Oregon | Hundreds of millions raised through 2025 | Oregon State University spinout |
| 1X Technologies | NEO | Home and light commercial | Norway / Silicon Valley | Hundreds of millions raised | OpenAI-backed |
| Foundation Robotics (Phantom) | Phantom MK1 | Defense, industrial | United States | Earlier-stage, less disclosed | Founder Sankaet Pathak |
| Sanctuary AI | Phoenix | General-purpose, dexterous tasks | Vancouver, Canada | Tens of millions raised | Geordie Rose (D-Wave co-founder) |
The most direct comparison is to Apptronik and its Apollo humanoid. Both companies are based in Texas, both draw heavily on NASA humanoid heritage (Apptronik through its actuator work on NASA Valkyrie at Johnson Space Center, Persona AI through Radford's role as Valkyrie's principal investigator at the same lab), and both pitch themselves to industrial customers rather than to consumers. The key difference is target task and stage. Apollo's first commercial deployments have been in automotive manufacturing with Mercedes-Benz and in logistics with GXO, where the work is fast, repetitive, and indoor. Persona AI is aiming at slower, heavier, more contact-rich work in shipyards and steel mills, where the value of a single successful weld or inspection is much higher and the tolerance for clean conditions is much lower.
The comparison to Figure AI is shaped by personal history. Jerry Pratt's tenure as Figure's CTO, and his departure to co-found Persona AI, was widely covered in the robotics press and remains one of the more public talent moves of the current humanoid cycle. Where Figure has pursued a high-burn, high-valuation strategy with marquee announcements about home humanoids and OpenAI partnerships, Persona AI is taking an explicitly capital-efficient industrial route. Whether that route turns out to be slower but more durable is one of the unresolved questions of the humanoid market.
The trade press treated Persona AI's emergence as a credibility event for the industrial humanoid thesis. Brian Heater's TechCrunch profile in June 2024 framed the company as one of the few new entrants whose founders did not have to build credibility from scratch.[2] The Robot Report's coverage of the pre-seed in May 2025 emphasized the unusual combination of a global shipbuilding customer and a heavyweight founder roster in a round that, by Series A standards, was still relatively small.[1] IEEE Spectrum's profile reinforced the same point and described Persona AI as one of the most experienced founding teams in the current humanoid wave.[7]
The HD Hyundai signing was widely covered in Korean and U.S. trade press, including The Korea Times, Robotics and Automation News, and AJOT, and was treated as evidence that the wave of humanoid pilots was moving beyond demonstrations into named programs with delivery dates.[5][16] POSCO's December 2025 investment was covered in Interesting Engineering and the POSCO Group newsroom and was interpreted as a second strategic vote from the Korean heavy-industry side.[6]
Independent commentary has tended to focus on two open questions. First, whether a smaller, deliberately scoped industrial humanoid program can compete on talent and capital with the much larger U.S. and Chinese humanoid efforts. Second, whether welding inside a shipyard, which is one of the harder real-world manipulation tasks available, is the right place to start, or whether Persona AI is taking on more difficulty than the current state of learned policies and contact-rich control can sustainably handle. The company's bet is that the same difficulty that makes welding hard also makes it valuable enough to pay for, and that starting with a hard task forces hardware and software discipline that easier tasks would not.[1][7]