Dialogue with OpenMind Founder: After receiving $20 million in investments from Pantera, Sequoia, and others, how far has the robot version of the "Android" system progressed?

Guest: Jan Liphardt, Founder of OpenMind

Interview organized by: momo, ChainCatcher

After decades of research and teaching at Stanford University and the University of California, Berkeley, Jan Liphardt, Associate Professor of Physics and Bioengineering, keenly observed that a profound structural transformation is occurring in the field of robotics.

On one hand, robots are accelerating from laboratories and factories into real-world scenarios, but their "brains" remain fragmented and closed. Over 150 hardware manufacturers are fighting their own battles, while mainstream software is still at the mechanical control level, making collaboration between systems difficult, and natural interaction between machines and humans, as well as value exchange between machines, impossible.

On the other hand, after more than a decade of development, blockchain technology, with its immutable ledger, decentralized governance, and micro-payment capabilities, has provided the infrastructural possibilities for identity recognition, trusted collaboration, and economic interaction between machines.

It is the intersection of these two trends that has given rise to Liphardt's core idea for founding OpenMind: to build an open and collaborative "operating system" for robots, akin to Android, enabling machines to possess cross-platform and cross-manufacturer social and collaborative capabilities, truly achieving thinking, learning, and cooperation.

In August 2025, this vision for OpenMind received a $20 million investment from several well-known institutions, including Pantera Capital, Sequoia China, and Coinbase Ventures.

With the completion of financing, the progress of OpenMind's core products and commercialization is accelerating. Its core open-source system OM1 has attracted thousands of developers worldwide, prioritizing technical integration with companies like Yushu Technology, Zhiyuan Robotics, UBTECH, Yujian Technology, Yundong Technology, Accelerated Evolution, Zhujidi Dynamics, and Zhongqing. Pilot projects are planned to gradually roll out in schools and home scenarios. The robot dog built on OM1 now has the ability to recognize its owner, remember items, and guard the home. Additionally, OpenMind is about to launch an application store specifically for quadruped and humanoid robots, with the first application already ready.

Recently, in an interview with ChainCatcher, Jan Liphardt elaborated on the motivations for entrepreneurship, technological pathways, and industry challenges, systematically analyzing the collaboration pain points in the robotics industry, the value of the open-source ecosystem, and the key role of decentralized protocols in achieving machine social collaboration.

Why do "robots" need a decentralized "Android"?

1. You have a deep background in bioengineering research and teaching at Stanford and other universities. We are curious, when and what prompted you to start focusing on blockchain and decentralized systems? How has this cross-disciplinary perspective reshaped your thinking about the future development of robotics technology, especially regarding the essence of intelligence and collaboration?

Jan Liphardt: We are indeed in a very special era where machines are gradually "awakening." I consider this from three perspectives:

First, artificial intelligence is inherently global. AI models do not care which country you are in. The speed of technological evolution is very fast, but establishing cross-national regulatory and governance frameworks often takes years or even decades. Questions like "Who is responsible?" or "Does this robot belong to the U.S., China, or Japan?" are ones we really don't have much time to dwell on. The decentralized nature of blockchain helps lower barriers, making international governance of AI and robotics more timely and effective.

Second, AI is not inherently good or rule-abiding. We must establish new systems to permanently record "what is real" and "what is right," which requires immutable ledger technology.

Third, like humans, non-biological thinking machines will also use economic logic to allocate their time and resources. Therefore, we must start building payment systems and market mechanisms suitable for them now.

Looking back, we are particularly fortunate. Over the past 18 years, thousands of people around the world have quietly built robust, secure, immutable, and decentralized ledger systems. It is likely that blockchain will become the core technological foundation for the coexistence and collaboration of 8 billion humans and increasingly intelligent machines in the future.

2. OpenMind aims to build a decentralized "Android" system for robots. How did this grand vision initially come about? How did you make the decision to transition from a stable academic career to the uncertain path of entrepreneurship?

Jan Liphardt: Our idea is quite straightforward. Today, there are billions of smartphones from different manufacturers, all running Android. The future of robots will be similar, so we also need an "Android" for thinking machines.

As for the transition from academia to entrepreneurship, my view is that life does not have absolute stability or security; every stage presents different opportunities and challenges. If you truly want to accomplish something significant, you must use different combinations of tools at different times. Sometimes you need to delve into basic science, and other times you need a well-funded, agile team with a commercial mindset.

3. What do you think is the biggest pain point currently facing the robotics industry? Why are traditional robotic system solutions unable to address these issues?

Jan Liphardt: In terms of hardware, the reliability of key components, such as dexterous hands, remains a bottleneck. Imagine a mechanical hand with five fingers and 12 degrees of freedom that fails after operating for just a hundred hours—what is its practical value? Of course, specific scenarios like logistics, transportation, defense, and surgery are maturing rapidly, but the industry is still exploring what the best entry point is for "general robots."

This field is currently very fragmented, with over 150 companies working on humanoid robots or other forms of robots. Many of these companies dream of becoming the "iPhone of robots," controlling all aspects of hardware, software, cloud, data, and ecosystem. However, we believe that, like the smartphone industry, the general robotics market will not have a single winner but will see many strong participants emerge.

On the other hand, traditional ROS-like software focuses on mechanical tasks and navigation, but what will make robots smarter and more useful in the future is often their social and cognitive abilities.

Moreover, ROS cannot provide hardware identity identification, built-in economic guarantees (such as deposits, penalties, or automatic settlements), nor does it have a convenient and secure payment mechanism, making it difficult for machines to prove their work in the real world to capital markets or DeFi (for example, as real-world assets or DePIN). In OpenMind's view, these capabilities are precisely key to achieving safety and high autonomy for robots.

4. Timing is crucial in entrepreneurship. Do you believe now is a critical juncture for realizing this vision? What technological maturities or market demand evolutions constitute this "timing"?

Jan Liphardt: Yes, we believe the time has come. Both robotic hardware and software have crossed the "good enough" threshold in many application scenarios.

Amazon has deployed over a million robots in its warehouses, and Waymo's self-driving cars are seen everywhere on the streets of San Francisco. We expect that by 2026, the first batch of general robots will enter American households and start doing genuinely useful tasks.

Of course, high-difficulty, precise actions like making ramen or slicing roast duck still pose challenges for humanoid robots. However, these are relatively marginal scenarios and should not be reasons to hinder robots from entering homes, schools, hospitals, and office environments.

On the other hand, in the field of cryptographic technology, foundational infrastructures such as stablecoins, Layer 2, and asset custody have also matured enough to support reliable and efficient micro-payment settlements between machines. Our machine settlement protocol in collaboration with Circle is a practical validation in this direction.

Collaborating with Chinese Robotics Manufacturers for Multi-Scenario Implementation

5. According to official information, OpenMind's current core products include the open-source robotic operating system OM1 and the decentralized network FABRIC protocol. What other important components are there? Can you systematically introduce the capabilities and divisions of the core products, and how they collaborate?

Jan Liphardt: In simple terms, the goal of OM1 is to make individual machines intelligent. FABRIC, on the other hand, is a global decentralized network primarily aimed at solving collaboration issues between machines and between humans and machines. For example, it allows robots to have their own identity identifiers, safely coordinate tasks with other machines or humans, and even support transactions of digital products between robots, such as certain robotic skill chips.

6. What is the current R&D progress of OM1? What capabilities need to be prioritized for breakthroughs next?

Jan Liphardt: Regarding the progress of OM1, there are several intuitive indicators. We have surpassed 2,500 stars on GitHub, while similar industry projects typically range from 30 to 80. We have over 300 active contributors, and in the last two weeks, about 7,500 independent developers have accessed our code repository.

In terms of implementation, OM1 has been adapted to various robot forms, including humanoid robots, bipedal robots, and robot dogs. It has also been used to drive the world's first machine dog governed by blockchain and connected to a decentralized network.

Our current work focuses on two directions:

First, we are starting to build customized models specifically to address the concrete bottlenecks encountered in real-world deployments; second, we are improving robotic simulation tools to better fit the development needs of "social robots," rather than just serving physical control. For example, current mainstream simulation environments like NVIDIA's Isaac Sim cannot simulate human voice interactions, which poses a significant barrier for social robots that need to converse with humans in environments filled with household noise, television sounds, children, and pets. We are investing efforts in these areas.

The robot equipped with OpenMind OM1 witnesses the launch of the first humanoid robot ETF, KraneShares KOID.

7. You have showcased robot dogs based on the OpenMind system at offline events and on social media. Can you share what capabilities these robot dogs currently possess? What advantages do they have compared to traditional system solutions?

Jan Liphardt: Our robot dogs currently possess numerous capabilities. They can recognize their owners, map the home environment, check surroundings and remember the locations of items like keys and glasses, and answer math questions and common queries. They also monitor their owners' status; for instance, if someone falls to the ground and remains motionless, the robot will assess whether medical assistance is needed and can help safeguard home security.

We are basically adding new skills to them every week. Current home tests generally run for 6 hours. The next important goal is to achieve 48 hours of continuous companionship, which means it must be able to autonomously recharge and learn to remain quiet at night without suddenly singing, speaking, or making noise, to avoid disturbing family members' rest.

8. We noticed that OpenMind is building a robot application store. Can you specifically introduce the vision and current progress of this application store?

Jan Liphardt: Yes, OpenMind is building an application store specifically for quadruped and humanoid robots. This allows users to download applications and skills for robots from a single platform, just like customizing phone functions through the Apple App Store or Google Play Store.

Last week, the first application was officially launched in the OpenMind application store. We are preparing marketing and developer education plans to attract thousands of developers worldwide to contribute new applications and skills, enabling quadruped and humanoid robots to create more value for users.

This application store is a "service layer" built on our core system, connecting developers with end-users. It serves as an important platform for discovering, distributing, and commercializing robotic skills, aiming to accelerate the prosperity of the entire ecosystem.

9. What is the current status of product delivery for OpenMind? What commercial progress or representative partners do you have?

Jan Liphardt: Our cloud infrastructure is being rapidly adopted, and we are collaborating with several Chinese robotics manufacturers to jointly design new products aimed at users in the U.S., Europe, and the Middle East.

For instance, we have listed companies such as Yushu Technology, Zhiyuan Robotics, UBTECH, Yujian Technology, Yundong Technology, Accelerated Evolution, Zhujidi Dynamics, and Zhongqing as our first batch of deeply integrated partners, planning to gradually roll out pilot projects in more schools and home scenarios in various regions.

By the first quarter of 2026, we expect to reach reference implementation agreements with these leading robotics manufacturers to promote the large-scale application of the ecosystem.

10. How can ordinary developers, robotics enthusiasts, and general users participate in or use the OpenMind ecosystem? Are there any corresponding incentive programs?

Jan Liphardt: Anyone can participate; just visit our GitHub page. We are also focusing on building organic developer adoption through our Developer Alliance Program (which offers $250,000 in credits to OM1 contributors), which has attracted over 100 contributors and more than 10,000 users building on the platform.

Participation in the OpenMind ecosystem is very open. Any interested developer can start by visiting our GitHub homepage. At the same time, we have launched the "Developer Alliance Program," which encourages contributions to OM1 through a total of $250,000 in credits. Currently, this program has attracted over 100 contributors, and thousands of developers are building on our platform.

What are OpenMind's Differentiated Advantages?

11. What is the current competitive landscape in this field? What are OpenMind's most distinct differentiated advantages compared to potential competitors?

Jan Liphardt: When comparing similar solutions in the industry, our positioning and distinctions are as follows:

For example, Husarnet, similar to FABRIC, also achieves collaboration between ROS2 nodes through peer-to-peer VPN. However, our difference lies in providing complete end-to-end encryption, more refined localization and network management capabilities, and crucially, we leverage blockchain to achieve network effects, encrypted payments, and secure model deployment, while not being limited to the ROS2 ecosystem.

Looking at Viam, which is a modular cloud robotics software platform similar to OM1, our focus is more on AI-native runtime and open-source technology stack, not just solving mechanical integration issues.

There are also research teams like Physical Intelligence that focus on foundational models for robots, concentrating on perception and manipulation. We explicitly consider this as one of the modules that can be integrated into OM1, rather than direct competition. We emphasize openness and transparency, deployment-oriented approaches, and the ability to flexibly invoke model capabilities from different sources at runtime.

Overall, our differentiation lies in becoming a neutral, modular cloud and coordination layer in the field of "embodied AI." We do not aim to compete with every emerging robotic application but rather provide an infrastructure that allows more hardware manufacturers, application developers, and researchers to access our technology stack. As the ecosystem expands, our models, tools, and data will form a positive feedback loop, and the prosperity of the ecosystem will reinforce our platform's value rather than dilute it.

12. What do you think is the most pressing challenge to overcome in the long journey of building a decentralized Android for robots?

Jan Liphardt: In the process of building a decentralized Android for robots, our current core challenges mainly focus on several technical aspects:

First, our evaluation system is still in a relatively early stage. Currently, we mainly rely on internally designed testing scenarios and use a "large model as the judge" approach to determine whether the robot truly understands instructions and whether its behavior is safe and reliable.

Second, the continuous adaptability after deployment is also a challenge. Each new home environment brings unexpected detail differences, such as lighting conditions, movies playing in the background, or specific tasks not covered in testing, all of which require the robot to adapt quickly. Additionally, how to smoothly connect and switch between high-level decision-making and low-level motion control is another area we are focusing on.

In terms of functional testing, we adopt a phased approach. New features are first validated internally through scripted tasks, log playback, and evaluation frameworks; then tested on controlled real hardware; and finally opened to select partners for pilot testing. Feedback from partners is crucial, as the metrics and opinions they provide from real application scenarios will directly feed back into our model iterations. At the same time, the integration of different forms of robots also helps us validate the system's compatibility.

These challenges are all part of our advancement plan, and the team is gradually tackling them.

13. What is your ultimate vision for the robotics industry or human-machine collaboration? Can you paint a specific scenario? What role will crypto ultimately play in this?

Jan Liphardt: OpenMind's long-term vision is a "social model." In the coming months to years, we will gradually build a foundational model for human-robot interaction based on existing cloud infrastructure and data capabilities, with privacy protection designed at the core of the technical architecture from the outset.

These models could become the "default brains" for the robots developed by our partners in the future. We anticipate that through these groundbreaking social and cognitive abilities, robots will help people learn more efficiently, receive more attentive care, and accomplish more valuable work in schools, hospitals, homes, and workplaces.

14. On the road to the "machine economy," what do you see as the biggest opportunities and potential risks (such as ethics, regulation, etc.)?

Jan Liphardt: When discussing the opportunities of the "machine economy," it is much like what we see in familiar sci-fi movies, where robots become trustworthy partners and assistants in life, with application scenarios that are almost limitless, from companionship in homes to industrial production.

However, potential risks also exist, as warned in these films. I sometimes find it incredible, for example, to see my 13-year-old child naturally asking a robot dog math questions. This makes us ponder: how can we ensure that this technology remains safe and benevolent?

At OpenMind, our answer is to ensure that the software running inside these machine brains is open and transparent, so that all humans can review, understand, and help fix issues when they arise.