(London) Can AI Agent Startups Survive Corporate Capture: Prof Ramesh Raskar, MIT
Friday August 21, 2026
10:00 AM – Noon
London UK
The AI-agent ecosystem is entering a decisive period.
Today, the foundational layers of the Agentic Web—discovery, identity, trust, payments, memory, coordination, and interoperability—remain open for invention. But this window may not remain open for long.
Over the next 18 months, major technology companies will attempt to turn their clouds, models, protocols, marketplaces, and agent frameworks into closed ecosystems. Startups that build directly on top of these platforms may gain early distribution—but risk becoming features, dependencies, or lowball acquisition targets once the market consolidates.
This session asks a practical question:
How can startups build enduring companies before the Agentic Web is captured by a handful of corporations?
Hear from Prof. Ramesh Raskar, founder of Project NANDA that originated at MIT and a pioneering architect of the Internet of AI Agents. For more than a decade, his work at MIT has explored the foundational technologies required for open, distributed, and trustworthy intelligent systems.
Prof. Raskar will share a strategic framework for founders navigating the coming infrastructure war.
What We Will Discuss
Why the next 18 months may determine who controls the Agentic Web
How corporate capture and ecosystem lock-in are likely to unfold
Why the protocol war may be a distraction from the infrastructure war
Which startup categories are most vulnerable to being commoditized
The danger of becoming a thin wrapper around a model, cloud, or agent framework
When partnering with a hyperscaler helps—and when it becomes a trap
Why agents must eventually operate across companies, clouds, models, and networks
Where open infrastructure can create durable startup opportunities
How startups can own trust, coordination, discovery, governance, and transaction layers
Why the long tail of specialized agents may ultimately be more valuable than a few universal assistants
How NANDA is an antidote to walled gardens (https://projectnanda.org)
Who Should Attend
This gathering is designed for:
Founders/Developers building AI-agent companies
Investors evaluating the emerging agent stack
Researchers working on multi-agent systems and the Agentic Web
Product and engineering leaders developing agent infrastructure
Entrepreneurs deciding where durable opportunities will emerge
Builders concerned about platform dependency and corporate lock-in
Schedule
10:00AM –10:30 AM — Check-in and Participant Intros
10:30AM – 11:15 AM — The 18-Month Window
Talk by Prof. Ramesh Raskar
11:15 AM – 11:30 AM — Discussion and Q&A
11:30 AM –12:00 PM — Networking
The Agentic Web can become an open ecosystem resembling the early internet—or a collection of corporate intranets controlled by a few dominant companies.
The architecture is being decided now.
Join us in London to discuss what startups should build before the window closes.
About Ramesh Raskar
Ramesh Raskar is an Associate Professor at the MIT Media Lab, where he directs the Camera Culture research group and leads the work on Networked AI Agents to build the "Internet of AI Agents." He is the founder of Project NANDA (Networked AI Agents in Decentralized Architecture), which is tackling the four infrastructure choke points of the agentic web: registry, credentials, interoperability, and attestation. His research vision holds that the next frontier of artificial intelligence lies not in ever larger centralized models but in a world of personal, edge based agents that people own, customize, and connect.
Beyond academia, he has worked on special research projects at Google [X], Facebook, and Apple, has co-founded and advised several startups, launched REDX.io as a platform for young innovators exploring AI for impact, and holds more than 100 US patents. He is widely recognised as one of the leading global voices on decentralized AI and the governance of the emerging agentic web, arguing that agentic commerce faces the same fork the World Wide Web once did, between closed walled gardens and a neutral, open infrastructure.
About Project NANDA
Project NANDA is building the foundational infrastructure for the Open Agentic Web, an open ecosystem where autonomous AI agents can securely discover one another, verify capabilities, communicate and collaborate across organizations and platforms.
Its work spans:
Agent discovery and interoperability
Trust, identity and verification
Agent-to-agent communication
Mobile and edge AI agents
Open standards for distributed intelligence
Civic and sovereign AI
Enterprise adoption of autonomous AI systems
The project is also developing NandaTown, an open developer sandbox where builders can experiment with real-world agent discovery, coordination and orchestration.
Research: https://nanda.mit.edu
Open Source: https://projectnanda.org
Developer Sandbox: https://nandatown.projectnanda.org
Join London Whatsapp channel: Join WhatsApp Group