Agentic Internet Workshop*

* An Internet Identity Workshop Inspired Event, Hosted by the IIW Foundation(IIW #41 is October 21-23)

The AgenticAI space is developing rapidly. Significant innovation is happening around protocols for how agents connect to each other, people, organizations, services, and things.

The goal of this one-day workshop is to bring individuals working on Agentic AI Protocols together to advance the field in a productive and generative manner.

What is the value of participation?

• Share your AgenticAI protocol work with others innovating protocols and get feedback.
• Learn who else is working on protocols, what are they working on
• Share what you are building with the protocols and discuss how to improve them.
• Gain a more in-depth understanding of key AgenticAI use-cases for protocols
• Discern if there is alignment for future work and collaboration.
• Consider how we can protect humanity, human integrity, judgment and creativity with these protocol stacks

The agenda will be set in the opening circle by the people who are gathered at the event using a process called Open Space Technology. Anyone in attendance can put a topic on the agenda.

The event builds on the legacy of the , a forum that has been meeting twice a year since 2005. In the first 10 years of the community’s life, we defined protocols such as OpenID Connect and OAuth, which are used a billion times a day on the internet today. In the second 10 years, we worked on decentralized identity protocols, like Decentralized IDentifiers, Verifiable Credentials, Trust Spanning Protocol, DIDComm, OpenID4VC and many others. Now, with this event, we want to focus on the agentic web and how personal and corporate agents can best engage with each other.

Notes will be collected in all sessions and made publicly available.

Places where AI & AIAgent Protocols are being worked on :

• AI Preferences (IETF AIPREF)
• Web Bot Authentication (BOF at IETF)
• AI Agents Side Meeting at last IETF
• AI agent applications in 6G network Side Meeting at IETF
• The Impact of AI Agent on Network Infrastructure Side Meeting at IETF
• AI Agent Protocol Community Group (CG) W3C
• AI KR W3C
• AI Identity Management CG OpenID Foundation
• AI and Trust WG Trust over IP
• Trusted AI Agents Decentralized Identity Foundaiton
• Linux Foundation Digital Trust
• A2A,
• MCP
• AGNTCY: A LF Project
• NLWeb
• KYAPay Protocol
• Agent to Agent Protocol (A2A)
• Agentic Profiles (A2A + DIDs)
• Decentralized Identifiers
• GNAP
• OAuth
• Trust Spanning Protocol

Tentative Schedule

8am doors open - breakfast and coffee

9am - 10am Agenda Creation in Opening Circle

10am - 11am Session 1

11am - 12pm Session 2

12 pm- 1pm Lunch

1pm - 2pm Session 3

2pm - 3pm Session 4

3pm - 4pm Session Summaries in Closing Circle

4:30pm post event social TBD

The Maximum Number of Attendees is 200

Cost:

We are asking attendees to submit a link of or description about their work on protocols that are relevant to AgenticAI.

• $150 for Independent Attendee
• $300 for Regular Attendee
• $1200 if you don’t have a submission about your current work in the field.

Event Hosts: Andor Kesselman, Kaliya Young

IIW Co-Founders: Phil Windley, Doc Searls

Potential Topics

The agenda will be co-created the day of the event by attendees. These are simply ideas folks have submitted as they register and shared to give perspective on what folks going into the event are thinking about discussing.

• Agentic Internet Registries
• The role of credential managers in the agentic AI ecosystem
• I'm interested in learning how / whether the web and browsers can help
• My goal for XMLUI is that a developer using agents to build XMLUI apps will enjoy a level of MCP support that reliably steers agents to correct patterns. I don't think this requires more or different protocols, it does require discovery and application of practical techniques. agentic experiences, and in what ways they may need to change.
• There is a ton of work going on in very different places. I'm hoping to leave this day with a better sense of who is doing what.
• The role of decentralized identity in personal AI agents.
• I would like to see how state management token are held in a secure way so that there is no privacy loss
• Particularly interested in delegation issues as well as dynamic authorization.
• MIT-NANDA
• why OAuth is not a fit for MCP
• I think I'm interested most in learning about MCP and how OAuth might relate to it, and specifically, how/whether browsers can help.
• DIDs and DIDCOMM are central to the overall Web 7.9 Agentic OS architecture.
• How to anchor agents to ground truth.
• The role of credential managers in the agentic AI ecosystem
• A2A protocol and cross app for Agentic AI
• AI permissions vs Human permissions
• Managing access rights
• The role of SSI in agentic identity and inter-agent protocols.
• The current state of standard and to see if an issue we see is being discussed or not
• ERC-8004, A2A, MCP
• The role of decentralized identity in personal AI agents.
• MCP MCP MCP MCP!
• MCP security
• A2A protocol and cross app for Agentic AI
• The role of decentralized identity in personal AI agents.
• agent to agent negotiation
• Access control to data
• Agents access control who are accesing on behalf of users
• How to incorporate an access layer and a payment layer in a decentralized way.
• trust & identity
• identity, trust in general for AgenticAI
• AI permissions vs Human permissions
• 1. How will cryptographic tokens represent delegation?
• 2. How is trust maintained across a network of downstream services?
• 3. How to express security policies to authorize access, enforce restrictions, and track obligations?
• 4. How will we govern agentic AI? What tools are needed that don't exist today? What are the limits of identity for governance?
• 5. What are the risks of delay while security and culture evolves to keep pace with existing progress?

Recommended Reading (we ask attendees when they register for suggestions)

• A Survey of AI Agent Protocols
• On Being Agentic
• Upgrade or Switch: Do We Need a Next-Gen Trusted Architecture for the Internet of AI Agents?
• I wrote a lot over the years at https://code.sgo.to around HTTP, APIs and identity. As far as agents are concerned, maybe my WS-REST keynote from 2014 might give a sense of what was on my mind at the time ( https://code.sgo.to/2014/09/05/ws-rest-2014-keynote.html). Don't think it needs to be broadly advertised though, likely out of date and incorrect.
• My vlog at collab101.org
• Bot or Not? Why Incentives Matter More Than Identity
• Roads, Robots, and Responsibility: Why Agentic AI Needs Identity Infrastructure
• AI Permissions vs. Human Permissions: What Really Changes?
• Articles here: https://gluufederation.medium.com/
• Draft OAuth AI Agents on Behalf of User
• The First Person Project White Paper at firstperson.network
• Email Verifications Protocol

Also check out other Meetups in Mountain View, Workshops in Mountain View, Business events in Mountain View.