IETF | Agentic AI communications: Identifying the standards we need

The IETF works on a broad range of networking technologies that provide the foundation for the Internet's growth and evolution.
While most IETF work takes place online, meetings and other events such as Hackathons provide in-person opportunities to advance work within the IETF.
The IETF is the premier Internet standards organization. It follows open and well-documented processes for setting these standards. Once published, those standards are made freely available.
The global IETF community works together in many different ways to produce high quality, relevant technical documents that influence the way people design, use, and manage the Internet.
By supporting the IETF today, you are ensuring a free and open Internet exists tomorrow.
In October 2026, the Internet Architecture Board (IAB) held a two-day workshop in Prague with people who are deploying post-quantum authentication. These are the first impressions from the co-chairs, ahead of the full workshop report.
Email is the oldest widely used application on the Internet, but it continues to evolve. At IETF 127 four working groups will meet to improve the security and usability of mail, and two Birds-of-a-Feather sessions propose improvements in mail operations and the way users sign up for mail subscriptions.
The IETF Administration LLC has prepared its draft budget for 2027 and now seeks community feedback.
These IETF 127 meeting sessions are likely to include discussions and proposals that are accessible to a broad range of Internet technologists whether they are new to the IETF or long-time participants.
Several new features implemented on RFC-Editor.org aim to make the site more useful and engaging for RFC consumers.
In October 2026, the Internet Architecture Board (IAB) held a two-day workshop in Prague with people who are deploying post-quantum authentication. These are the first impressions from the co-chairs, ahead of the full workshop report.
Email is the oldest widely used application on the Internet, but it continues to evolve. At IETF 127 four working groups will meet to improve the security and usability of mail, and two Birds-of-a-Feather sessions propose improvements in mail operations and the way users sign up for mail subscriptions.
The IETF Administration LLC has prepared its draft budget for 2027 and now seeks community feedback.
These IETF 127 meeting sessions are likely to include discussions and proposals that are accessible to a broad range of Internet technologists whether they are new to the IETF or long-time participants.
Several new features implemented on RFC-Editor.org aim to make the site more useful and engaging for RFC consumers.
When it comes to standards work around agentic AI, we’re at an exciting threshold. As more tools emerge, we’re seeing the amazing things it can accomplish. Now, we’re trying to figure out what parts of it need to be standardized.
There’s lots of work going on in this space, particularly at the IETF. But first, I want to identify a few of the processes and problems that might merit standardization when it comes to multi-vendor interoperability of applications over the Internet. (Remember, these are just my opinions on the subject.)
During IETF 124 in Montreal, Jonathan Rosenberg and I held a side meeting on this topic. You can view the strawman charter created for a potential working group.
With agentic AI (i.e. systems that allow AI agents to act on a human’s behalf), there’s a lot of communication going on: There’s human-to-agent, agent-to-tools, agent-to-agent. The human-to-agent communication is often very specific to that particular agent and how it wants to interact, so I don’t see a need for too much standardization there. But I think agent-to-agent or agent-to-tool communication will need it, because this often involves agents from one organization interacting with agents from another organization.
Here are some questions to consider:
Let’s say you’re planning your week at IETF 125, and you ask a chatbot (typing) to tell you the best restaurants in Shenzhen, and that chatbot responds with a list. As you continue to communicate with the chatbot to narrow down where you want to go, it will probably be easier to just switch to talking (audio) or perhaps look at photos from the restaurants.
This presents the need for multimodal capabilities with these tools. You might need images of a city map to know your proximity to the restaurants. And as the tool continues the conversation, it’s going to need to hold onto any context you’ve given it, whether that’s audio, text, images, Smell-O-Vision, etc.
Now say you’ve figured out where you want to have dinner, and you want the agent to make a reservation for you. It’ll need to now work with another agent, like one from OpenTable, to make the booking. That agent will need your credentials to make a reservation on your behalf, so how will it have access? And how can you limit what it knows about you? If it uses a 2FA system such as Visa Secure to access your credit card, how will that work?
The two AI agents can work together to choose a date, a time, and maybe a seating preference. This is where we need to keep humans in the loop. You can say yes, you want outdoor seating so you can enjoy a view of Shenzhen Bay. Or you can abort the mission altogether when it tells you the only availability is two weeks after IETF. Also, will you have to pay to make the reservation? If so, you will need to confirm this. (This opens the door to the topic of prompt fatigue. Will standardizing generate this among end-users?)
Our side meeting covered a lot of these topics, and even though we’re just at the beginning, we had 125 people in the room with roughly the same amount online, and we discussed a bunch of this work. Over the next few months, we’ll craft a proposed charter for an IETF working group that identifies some of the building blocks that need to be standardized to accelerate the full potential of agents. We will work with the creators of tod