13:48:39 RRSAgent has joined #adapt 13:48:43 logging to https://www.w3.org/2025/08/26-adapt-irc 13:48:43 RRSAgent, make logs Public 13:48:44 please title this meeting ("meeting: ..."), janina 13:49:03 Meeting: WAI-Adapt Teleconference 13:49:10 Chair: Lionel_Wolberger 13:49:17 Date: 26 Aug 2025 14:01:18 matatk has joined #adapt 14:02:05 Abhinav has joined #adapt 14:02:16 present+ 14:03:12 Lionel_Wolberger has joined #Adapt 14:03:14 present+ 14:03:24 present+ 14:03:25 present+ 14:03:29 scribe+ 14:05:51 regrets+ Annalu 14:05:57 scribe+ 14:06:40 rrsagent, make minutes 14:06:41 I have made the request to generate https://www.w3.org/2025/08/26-adapt-minutes.html Lionel_Wolberger 14:07:24 Topic: Symbols 14:07:53 matatk: I am ready to move the Symbols explainer forward this week 14:08:00 Lionel_Wolberger: We'll try and get you some time back for that 14:08:18 Topic: Destinations 14:10:40 General progress update: https://lists.w3.org/Archives/Public/public-adapt/2025Aug/0004.html - I didn't mention in here that I also updated the extension to Google's new Manifest V3 API, as well as bumping the packages. 14:10:46 rrsagent, make minutes 14:10:47 I have made the request to generate https://www.w3.org/2025/08/26-adapt-minutes.html matatk 14:16:20 Lionel_Wolberger: Is look and feel the same? 14:17:58 matatk: May go with sidebar rather than popup. 14:20:04 Topic: Well-known destinations and machine-readable semantics 14:20:11 rrsagnet, make minutes 14:21:10 Abhinav: I am considering how agentic AI would interpret our destinations 14:21:14 https://en.wikipedia.org/wiki/Agentic_AI 14:21:56 ... in my experience with recent LLM models and their reasoning 14:22:32 ... take for example a flight booking: I have some conditions, an overall travel time and a date range 14:23:00 ... an LLM can parse that out into source, destination, date range and constraints 14:23:31 ... it likely has a toolkit that it then draws upon, for example an API to a flight information service 14:23:48 ... and then another tool that takes that data, compares and filters as to the conditions it received 14:24:09 ... the filtering may be iterative: one filter for price range, one for flight duration 14:24:31 ... followed by another layer that may be able to use my credit card to book it 14:25:04 q+ 14:25:45 janina: *mentions https://innosearch.ai * 14:25:52 janina: Innosearch is an example of such a service, it uses expedia on its backend 14:27:11 ack matatk 14:27:46 matatk: While we will likely implement semantics to aid in machine readability 14:28:04 ... but I note that visual LLMs seem to be doing a reasonable job without semantics 14:28:41 Lionel_Wolberger: These visual presentations are coherent, typically thanks to underlying semantics 14:28:54 s/seem to be doing a reasonable/are available to do the/ 14:29:18 s/job without semantics/job without semantics - i.e. more like browsing sites visually/ 14:29:43 matatk: MCP that Abhinav mentioned: https://modelcontextprotocol.io/docs/getting-started/intro 14:29:50 scribe+ 14:30:36 Abhinav: Model context protocol was put forward by Anthropic, and is commonly used by these types of agents 14:31:10 ... it is a method to expose your LLM to such external services 14:32:08 ... it enables an Agent to select the proper tool. The MCP exposes the tools to the Agents 14:36:26 ... the previous generation of web agents, that use APIs to a website, can also use this MCP 14:36:55 ... and these web agents can be called via MCP 14:38:24 ... the MCP requires authorization 14:38:46 ... Our approach to destinations is very much required in this context. 14:38:59 ... Consider (a) AI Agents (b) tools (c) MCP interface 14:39:26 ... these agents are struggling to find the relevant pages, via DOM parsing or visual inspection to find the navigation 14:40:04 ... In my example, if a WKD clarifies that a website has an endpoint for flight search, the agent will go to that endpoint to get its results 14:40:12 rrsagent, make minutes 14:40:13 I have made the request to generate https://www.w3.org/2025/08/26-adapt-minutes.html Lionel_Wolberger 14:42:07 q+ to suggest we tell W3C payment processing about our linkset approach 14:43:40 q+ 14:45:00 janina: We should chase this at TPAC. 14:45:35 Abhinav: In summary, this WKD concept is supporting the current transition to Generative AI optimization 14:47:09 janina: Because we are making the URLs declarative 14:49:25 janina: Accessibility is more and more important, as accessibility relies on semantics and regularities that also benefit the LLMs 14:54:36 matatk: The 'destination' is the linkset 14:54:45 ... checkout is more of an action than a destination 14:54:53 ... it's a thing that you do, rather than a place that you go 14:55:08 ... Abhinav explained this as dependent on whether you have something 14:55:10 q+ 14:55:22 ... so we may need to revisit action v destination 14:55:35 ... also, janina brought up payments 14:55:46 ... there's payment for the final item, what you are buying; there's also paying for API usage 14:55:52 ... web payments is for the former not the latter 14:56:56 Lionel_Wolberger: I like it that 'actions' would be useful for LLMs. that could motivate us to tackle that next 14:57:27 Abhinav: Should we add this LLM agentic consideration to the explainer? 14:57:30 matatk: +1 14:57:33 Lionel_Wolberger: +1 14:57:52 matatk: A short addition to the use cases section is a good place to start 14:59:34 zakim, end meeting 14:59:34 As of this point the attendees have been Abhinav, matatk, Lionel_Wolberger, janina 14:59:36 RRSAgent, please draft minutes 14:59:38 I have made the request to generate https://www.w3.org/2025/08/26-adapt-minutes.html Zakim 14:59:44 I am happy to have been of service, Lionel_Wolberger; please remember to excuse RRSAgent. Goodbye 14:59:44 Zakim has left #adapt