The Summit That Changed Everything: Inside Project NANDA's Historic July Gathering
500+ visionaries, groundbreaking demos, and a roadmap to the future—here's what happened at MIT and how you can be part of what comes next
On July 14-15, 2024, something extraordinary happened at MIT. Over 500 researchers, entrepreneurs, technologists, and visionaries gathered for what may be remembered as the moment when the Internet of AI Agents moved from ambitious vision to serious research initiative. The Project NANDA: Architecting the Internet of AI Agents Summit wasn't just another tech conference—it was a convergence of brilliant minds united by a shared belief that the future of AI lies not in isolated systems, but in seamless collaboration across every platform, protocol, and organizational boundary.
For those of us who attended, the energy was palpable. This wasn't just about showcasing theoretical concepts or networking with industry peers. This was about witnessing the early stages of research that could fundamentally reshape how intelligence operates in the digital world. The presentations weren't just impressive—they were proof that the seemingly impossible vision of universal AI agent collaboration, while still evolving, has solid research foundations and growing momentum.
The Gathering of Minds: Who Came and Why It Mattered
The diversity of attendees at the NANDA Summit told the story of a movement that transcends traditional industry boundaries. Academic researchers sat alongside startup founders. Corporate executives from major tech companies engaged in deep technical discussions with independent developers. Government representatives explored policy implications with venture capitalists who were already calculating the market potential.
The speaker lineup read like a who's who of AI and decentralized technology innovation. Ken Huang, CISSP , CEO of DistributedApps.ai and a distinguished expert in AI and Web3, brought his extensive experience as Co-Chair of AI Safety Working Groups at the Cloud Security Alliance and his role as a core contributor to OWASP's Top 10 Risks for LLM Applications. His perspective on the intersection of AI safety and decentralized systems provided crucial context for understanding NANDA's governance challenges.
Rekha Singhal Head TCS Researcher leading research at TCS Paceport in New York, represented the enterprise perspective with her deep expertise in accelerating modern AI applications in multi-cloud environments. As an ACM Senior Member with over 100 publications and a Ph.D. from IIT Delhi, her insights into the practical challenges of deploying AI systems at scale resonated throughout the summit discussions.
The venture capital community was strongly represented by Patrick Chang from Dispersion Capital Capital, whose nearly two decades of VC experience and pioneering work in institutional web3 investment brought crucial insights into the funding landscape for decentralized AI infrastructure. John Werner from Link Ventures, known for his ecosystem-building expertise and leadership in innovation initiatives at Massachusetts Institute of Technology MIT Media Lab AHA at MIT Media Lab , provided valuable perspectives on how to scale the NANDA community effectively.
Gregory Raiz from FoundersEdge brought his unique approach to early-stage investing, which harnesses AI and deep understanding of founder psychology to identify promising entrepreneurs. His presence highlighted the growing investor interest in the agentic economy and the infrastructure needed to support it.
Sichao Wang from Cisco brought crucial expertise in security, risk, and fraud detection to the summit discussions. As a senior product management executive with experience at companies like Uber and Qualys, Wang specializes in building security and fraud detection products from inception. His insights into enterprise security challenges were particularly valuable for understanding how universal agent collaboration must address security concerns at scale.
Claudionor Coelho Jr , Chief AI Officer at Zscaler , provided high-level perspective on the intersection of AI and cybersecurity infrastructure. An ex-Googler with a background as both researcher and investor, Coelho's expertise spans the full technology stack from machine learning and software systems down to the fundamental semiconductor level. His understanding of how AI systems integrate with enterprise security infrastructure was crucial for discussions about deploying agent collaboration in enterprise environments.
From the enterprise technology sector, Victor Fong , a Senior Distinguished Engineer at Dell Technologies Partner Dell Technologies , delivered a keynote presentation on the NANDA Index Demo. His expertise in quantum computing orchestration and next-generation security services provided technical depth to discussions about the infrastructure requirements for large-scale agent networks.
Diane C. from Samsung Next Ventures brought her unique perspective shaped by her background as a video content director, writer, and producer, focusing on technologies that turn science fiction into reality—particularly AI, DeFi, Human-Computer Interaction, and longevity research.
Perhaps most significantly, Mahesh Lambe , the seasoned tech entrepreneur and AI architect who founded Unify Dynamics , served as a keynote speaker. With over 20 years of experience building large-scale systems and leading four startups (one acquired), Lambe's insights into the practical challenges of creating unified platforms that free enterprises from vendor lock-in directly aligned with NANDA's mission.
John Zinky , who started his career as a performance analyst for Arpanet Solutions (the precursor to the Internet), brought historical perspective and deep technical expertise in network performance optimization. As a principal performance analyst at Akamai AI , his insights into CDN and cloud infrastructure optimization were invaluable for understanding the scalability challenges facing universal agent networks.
But perhaps most importantly, the summit attracted hundreds of developers, researchers, and entrepreneurs who weren't representing major corporations or institutions—they were there because they recognized that this technology could democratize access to advanced AI capabilities and enable innovations that would be impossible within the constraints of current proprietary systems.
Conceptual Demonstrations: Research Vision Takes Shape
The technical presentations at the summit weren't operational systems—they were compelling research demonstrations and conceptual frameworks that showed how universal agent collaboration could solve real-world problems. Each presentation illustrated a different aspect of the theoretical infrastructure and its potential applications.
James Chen , CTO at Wonder , presented a compelling vision of how the NANDA infrastructure could transform food delivery operations. Wonder, the $7 billion food delivery and restaurant technology startup founded by Marc Lore, has been revolutionizing the restaurant industry with AI-powered technology. Their tech-enabled food halls feature 30+ virtual restaurant concepts under one roof, using AI for meal recommendations, nutrition profiling, and smart restaurant operations.
Chen's presentation focused on Wonder's core challenge: delivering hot food directly to customers' doorsteps. He outlined how, in a future enabled by NANDA infrastructure, kitchen timing agents could coordinate food preparation across multiple restaurant concepts, while logistics agents optimized delivery routes and timing to ensure food stays hot. Customer interface agents would manage real-time updates and delivery expectations, while temperature monitoring agents tracked food quality throughout the entire delivery chain.
What made this presentation so compelling wasn't operational technology—it was the clear articulation of how universal adapter infrastructure could enable these specialized AI systems to work together as if they were designed as a single, integrated system, once the theoretical framework becomes reality.
Maria Gorskikh presented her research on the NANDA Agent Adapter SDK concept—the theoretical software development kit that could enable developers to build applications on universal adapter infrastructure (see her technical presentation). Her presentation on the vision for connecting agents across cloud platforms like Azure and AWS showcased the potential applicability of NANDA's research direction. The SDK concept demonstrated how developers might eventually create agents that seamlessly communicate across different cloud environments without needing to understand the underlying protocol differences.
Working alongside Maria Gorskikh in the AI for Applications working group has provided me firsthand insight into the theoretical challenges that enterprises would face when trying to integrate AI agents across different platforms and organizational boundaries. Her SDK research addresses these challenges conceptually, exploring how developers might eventually have tools that abstract away the complexity of cross-platform communication while maintaining security and performance.
Pradyumna Chari serves as a Postdoctoral Associate at AHA at MIT Media Lab , specializing in decentralized AI, computer vision, and the Internet of AI agents. A prolific researcher, he has published in premier conferences like CVPR and SIGGRAPH and received accolades including a Best Demo Honorable Mention. Dr. Chari earned his Ph.D. from Indian Institute of Technology, Madras as the President's Gold Medalist, bringing both technical brilliance and academic excellence to the project's research foundations.
Ayush Chopra from Massachusetts Institute of Technology presented the theoretical foundations of the adapter system concept, outlining how agents might establish handshakes and communicate effectively even when they use completely different communication protocols. His presentation explored the research architecture that could make universal agent collaboration possible, including the cryptographic protocols for secure communication, the reputation systems for establishing trust, and the governance mechanisms for maintaining network integrity.
Chris Pease and Raj Simhan round out the core leadership team, bringing additional expertise in system design, business development, and community building. Their combined experience ensures that Project NANDA isn't just technically sound, but also practically implementable and commercially sustainable
Other presentations at the summit explored various applications and research directions for universal agent collaboration, demonstrating the breadth of potential use cases and the growing research community around these concepts.
A particularly significant development announced at the summit was the collaboration between Project NANDA and Radius , led by Robert Bench , Founder & CEO of Radius. Radius is building the payment stack for the agentic economy, and their first developer program has begun with NANDA, representing the largest agentic development community led by Ramesh Raskar . This partnership highlights the growing recognition that universal agent collaboration will require sophisticated payment and economic coordination mechanisms, and that the infrastructure for the agentic economy is already beginning to take shape.
Working Groups: Where the Real Work Happened
While the main stage demonstrations captured attention, the real substance of the summit happened in the intensive working group sessions. These collaborative workshops brought together experts from different domains to tackle specific challenges and explore concrete applications of the NANDA infrastructure.
The Web3 and Blockchain Integration working group explored how decentralized agent networks could leverage blockchain technology for trust, governance, and economic coordination. Participants discussed everything from token-based incentive mechanisms for agent collaboration to decentralized autonomous organizations (DAOs) that could govern agent networks.
The Agent Development and Communication Protocols working group dove deep into the technical challenges of building robust, scalable agent communication systems. Discussions covered protocol design, security considerations, performance optimization, and the practical challenges of implementing universal adapters across different platforms.
The Financial Applications working group, where significant exploration of NANDA's potential took place, examined how universal agent collaboration could transform everything from algorithmic trading to insurance underwriting to personal financial management. The group explored scenarios where financial AI agents could collaborate across institutional boundaries while maintaining the privacy and regulatory compliance required in financial services.
These working group sessions demonstrated how universal adapter infrastructure can create new business models and transform existing industries, with participants exploring specific prospects for applying NANDA's universal connectivity principles to real-world challenges across finance, technology, and decentralized systems.
The Enterprise Adoption working group focused on the practical challenges of implementing agent collaboration within large organizations. Participants from consulting firms and enterprise technology companies shared insights about the organizational, technical, and cultural barriers to AI adoption, and how universal agent infrastructure could address these challenges.
Research Progress: The Theoretical Framework Takes Shape
The summit wasn't just about presentations and discussions—it was also about sharing significant research progress that brings the Internet of AI Agents closer to theoretical viability.
The NANDA Index concept reached a major research milestone with the theoretical framework for distributed registry systems. This conceptual system would allow agents to publish their capabilities and discover other agents with complementary skills, functioning like a DNS system for AI agents. Unlike centralized directories controlled by individual companies, the NANDA Index is being designed as a theoretical framework that could be owned and governed by the global community of users.
The Universal Agent Adapter research achieved significant progress in theoretical protocol translation between major agent communication standards, including Anthropic's Model Context Protocol (MCP), Google's Agent-to-Agent (A2A) framework, traditional HTTPS, and emerging Natural Language Web (NLWeb) protocols. This research breakthrough suggests that agents built on completely different platforms could theoretically communicate seamlessly without requiring changes to their underlying architecture.
The X42 Payments research explored how agents might conduct secure financial transactions across different platforms and currencies, including both traditional payment systems and cryptocurrency networks. This theoretical capability would be crucial for enabling agent-to-agent commerce and creating economic incentives for collaboration.
Agent Pack Schemas were introduced as a theoretical standardized way for agents to describe their capabilities, requirements, and interfaces. These schemas could enable automatic discovery and integration, allowing agents to find and work with other agents without human intervention.
The Trust and Reputation research showed significant progress in theoretical frameworks for enabling agents to establish trust relationships even when they've never interacted before. The research suggests systems that could propagate trust through networks of agent relationships, creating a web of trust that could scale to billions of agents while maintaining security and reliability.
Global Collaboration: Building a Movement
One of the most remarkable aspects of the summit was the announcement of expanded global collaboration. Project NANDA now involves active participation from over 18 research institutions across six continents, representing a truly global effort to build the Internet of AI Agents.
This international collaboration isn't just symbolic—it's essential for ensuring that the infrastructure serves the needs of diverse communities and use cases around the world. Different regions face different challenges, have different regulatory environments, and have different cultural approaches to technology adoption. By involving researchers and developers from around the world in the design and development process, NANDA is building infrastructure that can truly serve everyone.
The global nature of the collaboration also provides resilience and prevents any single organization or country from controlling the infrastructure. This distributed development model ensures that the Internet of AI Agents will remain open and accessible, rather than becoming dominated by a few powerful entities.
Industry Impact: The Ecosystem Responds
The response from industry participants at the summit made it clear that Project NANDA has moved beyond academic research into serious commercial consideration. Major technology companies are beginning to integrate NANDA protocols into their development roadmaps. Startups are building business models around universal agent collaboration. Venture capital firms are actively seeking investment opportunities in the NANDA ecosystem.
The enterprise adoption discussions revealed significant pent-up demand for agent collaboration capabilities. Many organizations have developed sophisticated AI systems for internal use, but they struggle to make these systems work together effectively, let alone collaborate with external partners. The universal adapter infrastructure promises to unlock enormous value by enabling these isolated AI systems to work together.
The startup ecosystem is particularly excited about the democratizing potential of universal agent collaboration. Small companies with innovative ideas will be able to build agents that can immediately collaborate with the entire ecosystem of existing agents, rather than having to build everything from scratch or negotiate individual integration partnerships.
The Path Forward: What Needs to Happen Next
The summit concluded with a clear roadmap for the next phase of development. While significant progress has been made, there's still substantial work ahead to make the Internet of AI Agents a reality.
Technical Development continues to be a priority. The core infrastructure needs to be hardened for production use, with improved performance, security, and reliability. The protocol specifications need to be finalized and standardized. Developer tools and documentation need to be created to make it easy for developers around the world to build on the NANDA infrastructure.
Ecosystem Building is crucial for achieving the network effects that will make universal agent collaboration truly valuable. This means attracting more developers to build agents, more organizations to deploy them, and more users to benefit from them. The summit generated significant momentum, but sustained effort will be required to build a thriving ecosystem.
Governance and Standards need to be established to ensure that the Internet of AI Agents remains open, fair, and beneficial to everyone. This includes technical standards for interoperability, governance mechanisms for making collective decisions, and policies for preventing abuse and ensuring security.
Regulatory Engagement is becoming increasingly important as governments around the world develop policies for AI governance. The NANDA team is actively engaging with policymakers to ensure that regulations support innovation while protecting users and society.
How You Can Join the Movement
The most exciting aspect of Project NANDA is that it's designed to be a community effort. The infrastructure is being built as open source, with transparent governance and inclusive participation. During the summit, Ramesh Raskar shared the official resources and engagement opportunities for those interested in joining the movement.
Official Project Resources:
•LinkedIn Company Page: Follow @ProjectNANDA for updates and announcements
•Events Calendar: Stay informed about upcoming events at Lu.Ma/nanda
•YouTube Channel: Watch technical presentations and demos at @ProjectNANDA
•GitHub Repositories: Access open-source code and contribute at github.com/projnanda
•Project Homepage: Learn more at projectnanda.org
•MIT Research Group: Explore academic research at nanda.mit.edu
Direct Engagement Opportunities:
•Create Your AI Agent: Get hands-on experience at joint39.org
•Discord Community: Join technical discussions at discord.gg/BxnPBEqd88
•Core Contributors Slack: Apply to join the core team at forms.gle/VxaFLh1Dy6hgBo5C9
For Developers: The NANDA Agent Adapter SDK is available for experimentation and development through the GitHub repositories. The project welcomes contributions to the core infrastructure, developer tools, documentation, and example applications. The Discord community and technical working groups provide opportunities to collaborate with leading researchers and engineers while influencing the direction of the technology.
For Researchers: The project offers numerous opportunities for academic collaboration through the MIT research group and the global network of 18+ participating institutions. Research areas include fundamental work on agent communication and trust systems, applied research on specific use cases, and interdisciplinary studies on the societal implications of universal agent collaboration.
For Entrepreneurs: The NANDA ecosystem creates opportunities for new business models and applications that would be impossible with current technology. The Radius Developer Fellowship for Project NANDA, announced at the summit, provides a concrete example of how startups can build on the universal adapter infrastructure to create agents that immediately have access to the entire ecosystem of collaborating agents.
For Enterprises: Organizations can begin experimenting with agent collaboration within their own systems through the partnership opportunities outlined on the MIT research site. The enterprise working groups provide forums for sharing experiences and best practices, while the "Host Registry" and "Connect Your API" options allow companies to integrate with the NANDA infrastructure.
For Investors: The NANDA ecosystem represents a fundamental shift in how AI systems will operate, creating investment opportunities across multiple sectors and applications. The presence of major VCs like Dispersion Capital, Link Ventures, Founders Edge, and Samsung Next at the summit demonstrates the serious commercial interest in this space.
For Policymakers: The project welcomes engagement with government officials and regulatory bodies through the official channels. The global nature of the collaboration, spanning 18+ institutions across six continents, provides opportunities for international cooperation on AI governance and ensuring the Internet of AI Agents develops in ways that serve the public interest.
For Everyone Else: Even if you're not a developer, researcher, or entrepreneur, you can contribute to the movement by following the project on LinkedIn and YouTube, learning about the technology, sharing information with others, and supporting organizations and policies that promote open, democratic development of AI infrastructure.
The Urgency of Now
The July summit made one thing crystal clear: the Internet of AI Agents is not a distant future possibility—it's being built right now. The technical foundations are in place, the global collaboration is active, and the momentum is building rapidly. But the decisions made in the next few years will determine whether this infrastructure serves everyone or just a privileged few.
The window for shaping this technology is open, but it won't remain open forever. As AI capabilities continue to advance and the economic value of agent collaboration becomes more apparent, there will be increasing pressure to build proprietary systems that benefit their creators rather than the global community.
Project NANDA represents our best chance to ensure that the Internet of AI Agents develops as open, democratic infrastructure that serves humanity's collective interests. But this outcome isn't guaranteed—it requires active participation from people who share this vision and are willing to work to make it reality.
The summit showed that this vision is achievable. The technology works, the global collaboration is real, and the potential benefits are enormous. What happens next depends on whether enough people recognize the importance of this moment and choose to be part of building the future rather than just watching it unfold.
The revolution is happening. The infrastructure is being built. The community is forming. The question isn't whether the Internet of AI Agents will exist—it's whether you'll be part of creating it.
This concludes our three-part series on Project NANDA. The future of AI collaboration is being written right now, and there's still time to be part of the story. The choice is yours.
Thank you Hemanth Reganti for being part of the NANDA movement We're just getting started, and it means a lot to have curious minds like you with us. - Follow our journey: Project NANDA: Architecting the Internet of AI Agents - Join upcoming events: lu.ma/nanda