Add regular meeting notes from 09.02.2026

Meetings/RegularMeetings/2026-02-09.md

@@ -0,0 +1,95 @@
+# WebAgents CG: Regular Meeting (Feb. 9, 2026)
+
+## Agenda
+
+   * Introduction
+       * Introduction of new participants (if any)
+       * Review minutes from the previous meeting
+       * General CG updates
+   * Continue discussion on Interop Report
+       * Profiles (see [https://github.com/w3c-cg/webagents/pull/113/changes)](https://github.com/w3c-cg/webagents/pull/113/changes))
+   * Ontology for AI Agents (presentation by Paola Di Maio)
+
+
+Interop Report - Latest editor's draft: 
+
+[https://deploy-preview-107--w3cwebagentscg.netlify.app/](https://deploy-preview-107--w3cwebagentscg.netlify.app/)
+
+## Participants
+
+   * Antoine Zimmermann
+   * Jérémy Lemée
+   * Gustavo Nardin
+   * Arthur Casals
+   * Jean-Paul Calbimonte
+   * Max Ilyn
+   * Danai Vachtsevanou
+   * Paola Di Maio
+   * Rem Collier
+   * Eric Drury
+   * Andrei Olaru
+   * Andrei Ciortea
+   * Stephen Cranefield
+   * Ege Korkan
+
+## Regrets
+
+   * None
+
+**Scribe**: Andrei Ciortea
+
+**Notes from previous meetings**:
+
+[https://github.com/w3c-cg/webagents/pull/115/files](https://github.com/w3c-cg/webagents/pull/115/files)
+
+## Meeting Notes
+
+New participants:
+
+Eric Drury - a digital trust and digital identity advisor, working in various sectors - telecom, national ID, supply chain… I’m also a member of Trust over IP (LFDT). I’m not always able to join these calls but I do follow the discussion and emails as much as I can. My interest in AI Agents is in helping to assess their ‘trustworthiness’.
+
+AZ: reviews the meeting notes from last time. There are not general CG updates. We continue with the discussion of profiles. Danai to take lead.
+
+DV: Last time we started the discussion of agent profiles. We pick up from A2A agent cards. The main abstraction is of agent skill, which allows agents to decide when they should collaborate with one another for the completion of tasks, which is the major abstraction of agent communication in A2A.
+
+DV: We can also find here capabilities, which are protocol-specific capabilities. These are defined by the A2A protocol and have to do with some communication patterns, such as streaming.
+
+DV: In ANP, a profile can link to external services and products that agents can navigate in order to reason based on top of external resources — and then decide how to interact. We go towards more open systems where agents can discover not only other agents, but any resources on the Web.
+
+DV: There is explicit support for natural language interfaces vs. structured interfaces. ANP also aims to support JSON-LD, so it aims to support SemWeb technologies.
+
+EK: I remember from the initial ANP presentation that these were OpenAPI descriptions. These are no longer just that, right?
+
+DV: I believe so. [did not catch the rest of the discussion] [It's using JSON RPC 2.0 as a protocol]
+
+DV: Moving to LMOS, LMOS directly builds on top of the WoT TD ontology for the description of agents as WoT Things. It uses a specific vocabulary, the LMOS vocabulary, to define agent-specific metadata. But everything related to interaction, including security schemes, comes from the existing WoT approach. For example, we have observable properties, actions, and events. They do define some additional action types. This probably goes more to the interaction section. But the major difference here is that they use a specific ontology for describing hypermedia controls. Links are also present here, but maybe less explicit focus on linkability to other documetns in this specification — at least compared to the ANP specification.
+
+DV: This would conclude the approaches currently in scope for agent profiles.
+
+EK: We should merge open PRs for the interoperability report.
+
+AZ: We now move on to a different topic we received via email this morning. Paola Di Maio will present a recording of an ontology she created for Agentic AI.
+
+Link to recording from Paola: [https://drive.google.com/file/d/1G6igFKX9C7B5FC1La\_3gO7DjFJsNQ1Ft/view?usp=sharing](https://drive.google.com/file/d/1G6igFKX9C7B5FC1La\_3gO7DjFJsNQ1Ft/view?usp=sharing) 
+
+Link to Figshare paper in Paola's presentation  [https://figshare.com/articles/journal\_contribution/AIAO\_AI\_Agent\_Ontology\_v1/31231783](https://figshare.com/articles/journal\_contribution/AIAO\_AI\_Agent\_Ontology\_v1/31231783)
+
+The two slide sets I commented on during the pres are not yet shared but could be made available on request
+
+Thank you for inviting me to share this work!
+
+AZ: Thank you, Paola. I see multiple points of alignment.
+
+AC: We need to better coordinate efforts across multiple groups. Among others, we need to define which ontologies are needed for webagents and this is something we could collaborate with the AI KR group on.
+
+AZ: We also did some work in recent years in a project on Hypermedia MAS with some of the CG participants. We should see how we can integrate.
+
+PD: It could be useful to present the ontology in a previous meeting and see how we can integrate.
+
+AC: We could try to schedule a call. It would be good to also involve Fabien Gandon in the discussion.
+
+JL: The core ontology module is here: [https://purl.org/hmas](https://purl.org/hmas)
+
+AZ: Should this be part of the interoperability TF?
+
+AC: It could be a separate task force or part of the AI KR CG work.