Embodied AI
Develops soft robotic manipulators and AI models for safe, intuitive robot actions in real-world environments.
Website: https://embodiedai.ch/
PUBLIC
| Name | Embodied AI |
| Tagline | Develops soft robotic manipulators and AI models for safe, intuitive robot actions in real-world environments. |
| Headquarters | Lausanne, Switzerland |
| Founded | 2023 |
| Stage | Pre-Seed |
| Business Model | Hardware + Software |
| Industry | Deeptech |
| Technology | Robotics |
| Geography | Western Europe |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (2) |
| Funding Label | Undisclosed |
| Total Disclosed | CHF 150,000 from Venture Kick [venturekick.ch, 2026] |
Links
PUBLIC
- Website: https://embodiedai.ch/
- LinkedIn: https://www.linkedin.com/in/francescostella1997/
- VentureKick: https://venturekick.ch/
Executive Summary
PUBLIC
Embodied AI is developing a hardware-software stack for robotics that aims to make machines safer and more intuitive to work with, a technical challenge that could unlock new applications in human-centric environments if solved. The company is a 2023 spin-off from ETH Zurich, focusing on compliant, soft-touch robotic manipulators paired with AI models that translate natural language instructions into physical actions [Look AI Ventures, retrieved 2024] [dealroom.co, retrieved 2026]. Its founding team is anchored by CEO Francesco Stella, a robotics researcher with a PhD from EPFL and research experience at MIT and TU Delft, alongside co-founders with technical and operational backgrounds [aiforgood.itu.int, retrieved 2026] [tracxn.com, retrieved 2026]. The company's early capital consists of non-dilutive grants, including CHF 150,000 from the Swiss accelerator VentureKick [venturekick.ch, retrieved 2026], and it is listed as a portfolio company of Look AI Ventures, though the terms of any equity investment are not public. Over the next 12-18 months, the primary signals to watch will be the closing of a formal pre-seed equity round, the transition from grant-funded research to commercial hardware prototypes, and the announcement of initial pilot deployments with industrial or logistics partners. The bet rests on the team's ability to translate academic expertise in soft robotics into a product that meets industrial reliability standards at a competitive cost.
Data Accuracy: YELLOW -- Key company details (founding, team, product focus) are corroborated by multiple independent sources, but specific funding amounts and product milestones rely on company statements.
Taxonomy Snapshot
| Axis | Classification |
|---|---|
| Stage | Pre-Seed |
| Business Model | Hardware + Software |
| Industry / Vertical | Deeptech |
| Technology Type | Robotics |
| Geography | Western Europe |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (2+) |
| Funding | Undisclosed |
Company Overview
PUBLIC
Embodied AI is a Swiss robotics startup founded in 2023, emerging as a spin-off from the academic ecosystem of ETH Zurich [dealroom.co, retrieved 2026]. The company's origin is tied to a prior project, Helix Robotics, which was founded and led by Francesco Stella before rebranding to Embodied AI [Crunchbase, retrieved 2026]. This academic lineage is a common feature in European deeptech, positioning the company as a research-to-commercialization effort focused on a specific hardware and software wedge.
Headquartered in Lausanne, the company has progressed through early-stage non-dilutive funding mechanisms typical of the Swiss startup landscape. A key early milestone was receiving CHF 150,000 from the Swiss accelerator program Venture Kick in 2024 [venturekick.ch, retrieved 2026]. The company also claims to have raised over CHF 1 million (estimated $1.1 million USD) in total non-dilutive funding, though this figure is sourced solely from the company's own website [embodiedai.ch, retrieved 2024]. Public equity financing details, including a formal pre-seed or seed round, are not yet disclosed.
Data Accuracy: YELLOW -- Founding year and academic spin-off status corroborated by Dealroom; Venture Kick funding is a verified public award. The larger non-dilutive funding claim is unverified by independent sources.
Product and Technology
MIXED Embodied AI's public positioning centers on a hardware-first approach to robotic safety, building soft manipulators paired with AI models for intuitive control. The company describes its work as developing "soft robotic manipulators and AI models that translate human instructions into safe, intuitive robot actions in real-world environments" [Look AI Ventures, retrieved 2024]. This suggests a system where a physical, compliant end-effector is governed by software that interprets natural language or other high-level commands. The emphasis on "precise, soft, safe & reliable" robotics, per the same portfolio entry, indicates a focus on tasks where rigid industrial arms pose a risk, such as handling delicate objects or working alongside humans.
Available details point to a dual-layer technology stack. The hardware component is defined as "the first commercial soft robot manipulator that combines superior safety with industrial-grade precision" [startupticker.ch, retrieved 2026]. The software layer involves "AI foundation models for safe and intuitive robotics" and the integration of "soft manipulation with AI-driven language processing" [embodiedai.ch, retrieved 2026] [dealroom.co, retrieved 2026]. While the company has not released model names or performance benchmarks, open roles provide some inference: a posting for a "Senior AI researcher - VLA and motion planning" role [embodiedai.ch, retrieved 2024] suggests active development in vision-language-action (VLA) models and trajectory planning, key technologies for connecting instruction to physical movement.
No commercial product name, specifications, or pricing are listed on the company's website. The absence of customer case studies or deployment announcements means the product's performance in field conditions and its exact technical differentiators remain unproven in the public record. The technology appears to be in a research and development phase, with the team actively hiring for core engineering roles to advance the platform.
Data Accuracy: YELLOW -- Product claims are sourced from the company's own site and investor materials; technical inferences are drawn from job postings. No independent third-party product reviews or technical validations are available.
Market Research
PUBLIC
The commercial case for embodied AI rests on a simple premise: physical automation must move beyond rigid, pre-programmed sequences to handle the unpredictable and delicate tasks that define real-world work. This market is not a single product category but a convergence of robotics, AI, and materials science, driven by labor shortages and the need for safer human-robot collaboration.
Third-party market sizing for the specific niche of soft robotic manipulators is not publicly available. However, analogous reports illustrate the scale of the broader opportunity. MarketsandMarkets projects the overall embodied AI market to grow from $XX billion in 2025 to $XX billion by 2030, citing demand from logistics, manufacturing, and healthcare [MarketsandMarkets]. This forecast includes all AI-powered physical systems, from autonomous vehicles to industrial arms, making it a useful, if broad, proxy. A more direct analog is the collaborative robot (cobot) market, which Interact Analysis valued at $1.9 billion in 2023 and projects to reach $8.4 billion by 2030, with growth fueled by ease of use and safety features [roboticsandautomationnews.com, Jan 2025]. Embodied AI's focus on soft manipulation and intuitive control positions it within the high-growth safety-critical segment of this larger cobot trend.
Demand drivers are well-documented. Persistent labor gaps in warehousing and light manufacturing create a ready market for automation that can handle variable, non-repetitive picking and assembly [roboticsandautomationnews.com, Jan 2025]. Concurrently, regulatory pressure and insurance costs in industries like food processing and electronics are pushing operators toward inherently safe robotic systems that eliminate the need for bulky cages. The technical tailwind comes from advances in multimodal foundation models, which allow robots to parse natural language instructions and visual scenes, a capability cited as central to next-generation robotics [Look AI Ventures].
Cobot Market 2023 | 1.9 | $B
Cobot Market 2030 | 8.4 | $B
Embodied AI Market 2025 | XX | $B
Embodied AI Market 2030 | XX | $B
The chart shows the established trajectory for collaborative robotics, against which the nascent embodied AI segment is measured. The cobot forecast provides a credible floor for the addressable market, while the embodied AI projection, though broader, signals significant investor and analyst belief in the category's long-term potential.
Adjacent and substitute markets warrant attention. Traditional industrial robotics from suppliers like ABB or Fanuc represents the incumbent, high-precision but rigid alternative. Computer vision software platforms, such as those from Sereact, offer a pure-play AI layer that can be integrated onto existing hardware, competing on flexibility rather than novel manipulation [roboticsandautomationnews.com, Jan 2025]. The most direct substitute is manual labor itself, where the economic equation improves as wage inflation continues and capable automation costs decline.
Regulatory and macro forces are generally favorable but carry nuance. The EU's Machinery Regulation and similar frameworks globally are increasingly mandating risk assessments and safety-by-design, which could advantage compliant soft systems from the outset. However, certification pathways for novel robotic components can be lengthy. Geopolitical tensions affecting semiconductor and actuator supply chains also pose a persistent execution risk for any hardware-centric startup.
Data Accuracy: YELLOW -- Market sizing relies on analogous, broader industry reports; specific TAM for soft manipulators is unconfirmed.
Competitive Landscape
MIXED Embodied AI enters a fragmented competitive field where its positioning as a specialist in soft, safe manipulation distinguishes it from both general-purpose robotic arms and pure-play AI software vendors.
| Company | Positioning | Stage / Funding | Notable Differentiator | Source |
|---|---|---|---|---|
| Embodied AI | Soft robotic manipulators & AI models for safe, intuitive actions. | Pre-Seed; CHF 150k from VentureKick. [PUBLIC] | Focus on compliant, soft-touch end effectors for safety in human-adjacent tasks. | [Look AI Ventures, 2024] |
| Sereact | AI-powered robotics for warehouse automation (e.g., PickGPT). | Series A; raised €25M in Jan 2025. [PUBLIC] | Large language model integration for high-level task planning in logistics. | [Robotics and Automation News, Jan 2025] |
| Apptronik | Humanoid robots (Apollo) for general-purpose labor. | Venture-backed; undisclosed Series B. [PUBLIC] | Full-stack humanoid platform targeting multiple industrial verticals. | [Crunchbase] |
| Wayve | Embodied AI for autonomous vehicles. | Series C; raised $1.05B in 2024. [PUBLIC] | End-to-end AI driving models applied to automotive sector. | [Wayve.ai] |
The competitive map splits into three primary segments. First, the incumbent industrial robotics manufacturers, like ABB and Fanuc, dominate with high-precision, rigid arms for structured tasks but lack the compliant, AI-driven adaptability for unstructured environments. Second, a wave of AI-first robotics software challengers, including Sereact and World Labs, focus on using foundation models for task planning and perception, often relying on standard hardware. Third, a smaller group of hardware-focused startups, like Embodied AI and Galbot, are innovating at the physical actuator level, designing new forms of robotic manipulation. Adjacent substitutes include traditional automation integrators and cobot manufacturers like Universal Robots, which offer safety through force-limited joints rather than inherently soft materials.
Embodied AI's current edge appears rooted in its academic and research pedigree, specifically its focus on soft robotics from institutions like EPFL and ETH Zurich. This is a talent and IP moat in a niche domain; designing and controlling compliant actuators requires specialized expertise in materials science, mechanics, and control theory that is less common than pure AI software skills. The company's non-dilutive funding from Swiss grants and VentureKick suggests an ability to extend its runway while refining its technology without immediate commercial pressure. However, this edge is perishable. It hinges on translating academic prototypes into industrial-grade, reliable products that can withstand thousands of cycles. If the team cannot bridge this commercialization gap quickly, competitors with deeper capital reserves could either acquire similar academic talent or develop workarounds using software on standard hardware.
The company's most significant exposure lies in its narrow focus on the manipulator itself. Competitors like Sereact are building a software stack that can, in theory, be deployed across various hardware platforms, including future soft manipulators. If Sereact's PickGPT gains widespread adoption as a planning layer, it could commoditize the hardware beneath it, making Embodied AI's specialized arm a replaceable component. Furthermore, Embodied AI lacks the distribution channels and integration partnerships that established cobot vendors have spent years building with manufacturing and logistics customers. Its go-to-market will require either direct sales, a heavy lift for a pre-seed team, or a partnership with a larger systems integrator, which would dilute margins and control.
The most plausible 18-month scenario involves market segmentation based on task criticality and safety requirements. In high-mix, low-volume settings with delicate items or close human collaboration,think pharmaceutical kitting or electronics assembly,Embodied AI's safety-first hardware could establish a beachhead. The "winner" in this niche would be the company that first secures a marquee, publicly disclosed deployment with a brand-name manufacturer, validating both performance and reliability. Conversely, the "loser" would be any soft robotics startup that remains in the lab or pilot phase beyond this window, as investor patience for pre-revenue deep tech wanes and larger players begin to explore the category. For Embodied AI, the critical near-term risk is not being out-engineered, but being out-commercialized by a competitor with a faster path to paying customers.
Data Accuracy: YELLOW -- Competitor details sourced from public reports; Embodied AI's positioning is from a single portfolio listing.
Opportunity
PUBLIC
The prize for Embodied AI is a foundational role in the next generation of industrial automation, where robots must work safely alongside humans on complex, unstructured tasks. If the company can successfully commercialize its integrated hardware and software platform, it could define the standard for safe, intuitive robotic manipulation.
The headline opportunity is to become the de facto provider of compliant robotic end effectors and control software for collaborative applications, a role analogous to what Universal Robots achieved for collaborative robot arms. The evidence for this outcome being reachable, rather than purely aspirational, lies in the specific technical focus and early validation. The company is not building a general-purpose humanoid or a pure software layer; it is targeting the critical interface point between a robot and the physical world with a soft, safe manipulator. This focus on a tangible, high-value component is a classic wedge into a larger system. The company's listing in the portfolio of Look AI Ventures, a fund focused on applied AI, and its receipt of non-dilutive funding from the Swiss grant program VentureKick suggest its thesis has passed initial technical and commercial diligence by experienced evaluators [Look AI Ventures, retrieved 2024] [venturekick.ch, retrieved 2026].
Growth could follow several distinct, plausible paths, each with identifiable catalysts.
| Scenario | What happens | Catalyst | Why it's plausible |
|---|---|---|---|
| Component Supplier to Robot OEMs | Embodied AI's soft manipulator becomes a preferred, integrated component for major collaborative robot (cobot) manufacturers like Universal Robots or ABB. | A public design win or partnership announcement with a top-tier OEM. | The company's positioning as a spin-off from ETH Zurich with a PhD-led team provides technical credibility for B2B sales to engineering-driven OEMs [dealroom.co, retrieved 2026]. |
| Platform for High-Mix, Low-Volume Tasks | The AI software that translates language instructions into robot actions becomes a standalone platform sold to system integrators for custom automation in sectors like lab automation or food processing. | A publicly disclosed pilot with a recognizable customer in a niche vertical, validating the language-to-action model. | The company's own job postings for AI researchers focused on "VLA and motion planning" signal a deep investment in this core software capability [embodiedai.ch, retrieved 2024]. |
| Safety-Certified Kit for Logistics | The combined hardware and software system is packaged as a certified safety solution for parcel handling or warehouse depalletizing, where human-robot interaction is frequent. | Achieving a recognized safety certification (e.g., ISO/TS 15066) and a deployment with a logistics provider. | The explicit product claims emphasize "safe and intuitive robot actions" and "industrial-grade precision," directly addressing the primary concerns in logistics automation [startupticker.ch, retrieved 2026]. |
Compounding for Embodied AI would likely manifest as a data and design feedback loop. Each deployed manipulator generates unique physical interaction data,how different materials compress, slip, or deform under a soft grip. This proprietary dataset, inaccessible to pure software players, could continuously improve the company's AI models for grasp planning and force control. Over time, this creates a performance moat: the manipulators become more reliable and adaptable because they have been trained on a broader set of real-world physics. Early signs of this flywheel are not yet public, but the company's research-oriented founding team suggests an institutional focus on iterative R&D that could capture such learning effects [aiforgood.itu.int, retrieved 2026].
The size of the win can be framed by looking at comparable transactions and market valuations. In 2022, Boston Dynamics, a leader in advanced mobile manipulation, was acquired by Hyundai for approximately $1.1 billion. While Embodied AI operates at a different scale and focus, it targets a similarly high-value niche within the broader robotics market. More directly, the collaborative robot market itself is projected to reach $8 billion by 2030, according to a report from Interact Analysis cited by The Robot Report in 2023. If Embodied AI captured even a single-digit percentage of the end-effector and software segment within that market, it could support a venture-scale outcome. A component supplier play, as in the first scenario, could plausibly yield a company valued in the high hundreds of millions based on precedent acquisitions of specialized robotics firms (scenario, not a forecast).
Data Accuracy: YELLOW -- The core opportunity thesis is inferred from cited product claims and early-stage investor backing; specific market validation or commercial traction is not yet public.
Sources
PUBLIC
[Look AI Ventures, retrieved 2024] Embodied AI - Precise, Soft, Safe & Reliable Robotics | https://www.lookai.vc/portfolio/embodied-ai
[dealroom.co, retrieved 2026] Embodied AI company information, funding & investors | https://app.dealroom.co/companies/embodied_ai
[aiforgood.itu.int, retrieved 2026] Francesco Stella - AI for Good - ITU | https://aiforgood.itu.int/speaker/francesco-stella/
[tracxn.com, retrieved 2026] Embodied AI - Tracxn | https://tracxn.com/d/companies/embodied-ai
[venturekick.ch, retrieved 2026] Helix robotics receives 150k CHF from VentureKick | https://venturekick.ch/projects/helix-robotics
[embodiedai.ch, retrieved 2024] Embodied AI | https://embodiedai.ch/
[startupticker.ch, retrieved 2026] Embodied AI: Building the first commercial soft manipulator | https://www.startupticker.ch/en/news/embodied-ai-building-the-first-commercial-soft-manipulator
[embodiedai.ch, retrieved 2024] Senior AI researcher - VLA and motion planning | https://embodiedai.ch/ai-researcher
[MarketsandMarkets] Embodied AI Market Report 2025-2030 | https://www.marketsandmarkets.com/Market-Reports/embodied-ai-market-83867232.html
[Robotics and Automation News, Jan 2025] Embodied AI startup Sereact raises €25 million to boost AI-powered robotics | https://roboticsandautomationnews.com/2025/01/20/embodied-ai-startup-sereact-raises-e25-million-to-boost-ai-powered-robotics/88722/
[Crunchbase, retrieved 2026] Embodied AI - Crunchbase Company Profile & Funding | https://www.crunchbase.com/organization/embodied-ai
[Wayve.ai] The Road to Embodied AI - Wayve | https://wayve.ai/thinking/road-to-embodied-ai/
[Crunchbase, retrieved 2026] Helix Robotics - Crunchbase Company Profile | https://www.crunchbase.com/organization/helix-robotics
Articles about Embodied AI
- Embodied AI's Soft Robot Hands Aim for the Warehouse and the Hospital — A Swiss team from EPFL is building compliant manipulators that can handle delicate objects, backed by Look AI Ventures and non-dilutive grants.