MIT Media Lab Professor, ProjectNanda for AgenticWeb, PathCheck Foundation, Femto-photography, Digital Health and AI, Impact Innovation
MIT Media Lab
Joined June 2009
- Tweets568
- Following296
- Followers4.8K
- Likes199
NANDA Town as a sandbox for experimenting with agent societies, protocols, and emergent behavior.
podcasts.apple.com/us/podcas…
Virtual Hackathon for Agentic Web
Join virtually in NandaHack, the online Agentic AI Hackathon from HCLTech and the MIT Media Lab, focused on building the infrastructure and services needed for an open Internet of AI Agents.
nandahack.media.mit.edu
linkedin.com/feed/update/urn…
🔸CM Devendra Fadnavis launched the app ‘Kumbhdoot: AI Agent for Every Pilgrimage’ at the ‘India AI Impact Summit 2026’ in New Delhi. The app has been developed to provide guidance and assistance to devotees attending the upcoming Simhastha Kumbh Mela in Nashik.
Minister Adv. Ashish Shelar, Minister Nitesh Rane, and other dignitaries were present.
🔸मुख्यमंत्री देवेंद्र फडणवीस यांनी नवी दिल्ली येथे आयोजित ‘इंडिया AI इम्पॅक्ट समिट 2026’मध्ये नाशिक येथे होणाऱ्या सिंहस्थ कुंभमेळ्यासाठी भाविकांना मार्गदर्शन व सहाय्य करण्यासाठी विकसित करण्यात आलेले ‘कुंभदूत : पवित्र यात्रेसाठी तुमचा AI साथीदार’ हे ॲप लॉंच केले.
यावेळी मंत्री ॲड. आशिष शेलार, मंत्री नितेश राणे व इतर मान्यवर उपस्थित होते.
@Dev_Fadnavis @ShelarAshish @NiteshNRane
#NewDelhi #DevendraFadnavis #IndiaAIImpactSummit2026 #AIForKumbh
This was a deep conversation about the 3 phases and what it means to have billions of personal agents and what kind of 'agent societies' will emerge
youtube.com/watch?v=6jTg_gSD…
At Project Nanda, this vision is foundational: privacy-first AI, individual agency, and open agent economies built into the infrastructure itself.
nanda.mit.edu
AI is moving from centralized “cloud intelligence” to personal, edge-based agents.
From foundry → garage → bazaar.
From renting intelligence to owning it.
From one “God model” to societies of interoperable agents.
or create your own agent at join39.org that comform to Nanda architecture
But there’s still work ahead:
= Standardized discovery and identity layers across silos
= Robust governance/ safety frameworks for open agent societies
= Open incentive/ economic protocols for agent commerce
What’s Real — and What’s Next
What we’ve seen so far is more than hype:
= Independent agents executing real tasks
= Emergent agent-to-agent interactions that resemble market formation and cultural patterns
1. Agent Foundations — proving the basic intelligence and primitives of agents.
2. Agent Commerce — agents engaging in economic activity and coordination.
3. Agent Societies — agents forming persistent, self-sustaining cultural, social, and institutional structures.
Ramesh Raskar retweeted
Finally! A reason to feel optimistic about AI?
There is a sense of doom and gloom at the moment with AI.
In step @raskarmit , @MIT professor...
What if AI could be democratized in a way that we could all own a piece of it...
Episode out now: linktw.in/cuLSfp
The NANDA Index is the Phase1 component of #ProjectNANDA's three-phase vision for building the Internet of AI Agents. Using a city analogy: if A2A protocols are the streets (that agents use to communicate), Project NANDA builds the entire city, NANDA Index is the address system.
Ramesh Raskar retweeted
An interesting AI hackathon is unfolding at MIT right now. Think Game of Thrones meets machine learning - where alliances and strategy matter as much as technical skills to build the best model.
Join the MelaSafety.org Reels challenge by creating a 3-minute video or a short booklet to create SAFETY AWARENESS videos. Turn MahaKumbh Pilgrims into Safety advocates. Your content could rewrite crowd psychology! Showcase your creativity, win cash prizes!
Ramesh Raskar retweeted
What next?
1. Know more: web.media.mit.edu/~ayushc/po…
2. Act?: Work with us @MIT and @UVA_BI to track these complex dynamics - from fowl to fork and stop the next pandemic!
@raskarmit @techreview @medialab