Submitter's message Please check the final agenda item, our discussion of AI and DITA; I tried to keep it very neutral, but for those of you with concerns, please check what I've recorded and let me know if anything needs to be revised. I was more direct about our discussion of the crazy AI-generated DITA books, but I think that's totally appropriate; we are, in fact, the experts...
Thx,
Nancy
ActionItems:
1. Stan will contact all the authors who've signed up to do CNS, and check in with them; so he can report back in Sept.
===============================================
Minutes of the OASIS DITA TC
Tuesday, 25 August 2026
Recorded by Nancy Harrison
link to agenda for this meeting:
https://github.com/oasis-tcs/dita/wiki/Previous-agendas
Attendance:
===========
Robert Anderson, Stan Doherty, Kristen Eberlein, Nancy Harrison, Bob Johnson, Zoe Lawson, Lwroy Steinbacher, Lwroy Steinbacher, Frank Wegmann
Business
========
Regrets: Christina Rothwell
1. Minutes
18 August 2026 (Harrison, 19 August 2026)
https://groups.oasis-open.org/discussion/dita-tc-minutes-18-august-2026-uploaded
Kris moved, 2nd Bob, approved by TC
2. Announcements; new items for agenda
- Stan; we're organizing Sept Boston UG mtg, a review of the crazy DITA books I've been reviewing; I found yet another one.
- Kris; this would be a good time for agenda item about new DITA implementations these days, wrt 'how is AI helping or hindering implementations? e.g. are companies not hiring consultants, because company execs think AI can do everything?
- Frank; yes, please, we do need to have that conversation.
- Robert; sounds like fun.
3. Review: 2025-2026 DITA Books uploaded
Stan Doherty, 21 August 2026
https://groups.oasis-open.org/discussion/review-2025-2026-dita-books-uploaded HTML5 version
https://modularwriting.com/dita-books/
- Stan; most of these are by folks with very little competence, and none of them are really any good, but some highlight things we need to think about in writing good ones. That is, they highlight the things we would like to see in a book. So, what are the common gaps? common strengths? What does this tell us about what would be important?
- Stan; basically, these books were created by putting in a bunch of AI prompts and just printing all the results of those prompts. If we did those AI prompts, we'd get that data, but because we actually have used DITA, we would massage the data to make it useful; most of the data you get from the set of AI prompts most of these guys seem to be using is redundant, so if it's not edited well, as it is not, for the most part, it's just not that helpful or easy to use.
- Kris; I'm interested to see that one did a strange mangling of authors and titles of existing DITA books
- Stan; also, beecause of the nature of AI, none of them make refs to the actual spec, and so they miss out on the single most complete DITA resource that exists.
- Kris; my thought, over the years, we've tried very hard to improve examples; they're not normative, but we've worked very hard on them so that they issustrate things people need to know. In future, could we take examples and move them towards tutorial direction for a CN?
- Robert; and we fixed all the examples for 2.0, so they're all correct., thanks to automated testing. :-)
- Kris; e.g., in writing examples about constraints and constraint examples, went to great lengths to make them real, substantive ones.
- Nancy; maybe it would be useful to have a CN of all our examples?
- Stan; could a library of examples be generated via automation?
- Robert; it could, but might not be so useful, it would be a bit of a jumble.
- Kris; it would require a bit of hand work. we don't have a lot of metadata in our spec material, so it might be hard to do it in a useful way. But please, folks, do think about looking at them and organizing them into a tutorial.
4. Committee notes effort (Doherty)
Committee Notes.
https://github.com/oasis-tcs/dita/wiki/DITA-TC-Committee-Notes Updated description of mission
https://groups.oasis-open.org/discussion/updated-mission-description
***ActionItem: Stan will contact all the authors, and check in with them; so he can report back in Sept
- Stan; fyi, Oliver is willing or reorganize the Github repo, but he doesn't have write permission.
- Robert; anyone can create a fork and put stuff in; they don't need write permission, You only need write permission for merging; which I do; after you do the work on your forked branch, you submit a pull request to get the work merged in.
- Stan; I'll ask Oliver to contact Robert and Kris about it.
- Robert; anyone can fork the spec repo and create a branch, no write permission needed, only required to merge back into main repo.
- Kris; there's an issue with way repo was set up. The main branch is only for CNs that have already been published; There should be both a dev and a publish branch, but there isn't.
- Stan; well, I'll send him to you
5. Update on spec progress (Anderson)
- Robert; I still haven't set up CF at SynchroSoft. I'm continuing to close out draft-comments work I started in summer, at this point, I've closed out all the draft-comments in conditional processing stuff, so we could review that first.
- Kris; we had a pre-review and it was clear there was a lot of work to do on it.
- Robert; but we did do the DITAVAL review
- Kris; right; I was better able to keep track of things in the old Kavi wiki; I'll set up a call with you to discuss
- Robert; ok
6. General discussion on AI
- Kris; so, let's talk about issues with AI?
- Frank; what kind of implementations are there as of now?
- Kris; there aren't specific implementations, but there are 3 general ways that companies decide to use DITA for their content.
- Kris; maybe our team isn;'t seeing a lot of new DITA implementations.
- Bob; I spent some time trying to generate interest among equip. mfrs, and didn't see interest in the strategic investment required to make a DITA migration. Everyone's tactically focussed on diff. economic circumstances; I've reached out about iiRDS but got same response; too much economic uncertainty right now to invest in major projects that don't have immediate payoff..
- Kris; what do you see, Leroy, with companies use Adobe with new DITA implementations? are they less traditional companies?
- Leroy; because Guides is integrated with other Adobe tech, we get companies that are using DITA in non-traditional - i.e., non techdoc - ways. So marketing guys say 'don't mention DITA. these aren't tech docs, but we have a need for content reuse' So even where I suggest DITA for content reuse, I always try not to put too much emphasis on DITA as DITA, because folks see it as a tech doc thing. But we use it with AEM customers; they have AEM, but need Guides.
- Kris; at Mayo, decision has come down that most writers can use Word for Guides content(careful about minutes) seeing a couple of things, but content is stored in DITA, so it's the content mgmt folks who end up using it.
Also, speed of publication is so critical with our content; that it takes precidence over reuse I.wonder if is this is happening with other companies. DITA's real value for us is in being able to use DITA-OT for omni-channel publishing.
- Leroy; we see the same thing 'casual content contributors' goal is to set up a pipeline where that content can be converted to structured, but folks are not authoring in it.
- Frank; sounds like 'casual content contributors' don't have the mindset of creating structured content, so it's hard to get them to write in a structured context?
- Leroy; it's 'the way they've always done it' but company wants to pull it into a repo and have it available multi-channel. so have to have a role of content mgmt/organization. AI is a huge part. customers want AI to create everything for them, e.g. metadata. XML is being taken over by AI, so they understand that 'structured content can be used better by AI', but it can be marginalized.
- Kris; leadership wants structured authoring for its benefits, but they don't understand the work involved in creating it. And there's a strong sense of 'AI magic.' I think we could have a future where, when a publication gets revised, all maps and images are thrown away and created new, because it's faster that way.
- Leroy; of course, it's an open question as to whether that actually saves money/time, or eventually cost more.
- Kris; another question is whether companies have actually explored how AI can help them with reuse.
- Lwroy; our partners are looking at that.
- Kris; AI should be perfect for helping with reuse, especially wrt conref'd content. if you're using agents to do a first draft of content, they could be aware of content that needs to be reused on all docs and provide that. don't knw what kind of explorations folks have been doing with AI besides the obvious ones.
close 11:49am EDT
-- Ms. Nancy Harrison
---------------------------------
Nancy Harrison
Principal, Infobridge Solutions
Nancy Harrison (Personal)
---------------------------------