@understandinggroup.bsky.social
Things are changing so fast that a good measure of the value of an article is whether it is still pertinent three months later. This article is about that old, and it definitely still applies.
19.02.2026 19:02 β π 0 π 0 π¬ 0 π 0
Going back to some of our "best of" posts from 2025, this excellent article on what an ontologist is, the work they do in collaboration with teams, and how to become one is one of our favorites.
buff.ly/jeDvqqu
Happy Valentines day to those who celebrate, from all of us at The Understanding Group! #TUG #lovewins
13.02.2026 18:04 β π 0 π 0 π¬ 0 π 0
We would say that "artifact" is a better term than "documents", because it describes something made that is durable and can be experience by other people. The most important thing however, understanding how an artifact will be used and understood to do work.
12.02.2026 18:03 β π 0 π 0 π¬ 0 π 0
This study proposes that AI will lower transaction costs of doing business between firms, to essentially zero.
We're not sure we like how they calculate "cost", but the paper shows how #AI can change how we think about all kinds of things.
We keep coming back to this profound insight from Brain-Food. This entry #649. Check out the entire site here:
buff.ly/kkfi4sM
Jesper's elegant description of an "architecture of agency" maps to a lot of the "behavioral architectures" in the world, such as urban planning, Information Architecture, and manufacturing design. Good stuff!
buff.ly/KT7x5f4
Spot-on from Ethan Mollick:
"If we donβt think hard about WHY we are doing work, and what work should look like, we are all going to drown in a wave of AI content..."
We think this post by Scott Jenson remains pretty accurate, with his two axioms about how technological innovation works:
- Stuff gets useful as it gets more widely available
- you have to leave a new innovation alone in the wild to figure out its actual value.
Mental Models for Prompts. Techniques like #InformationArchitecture more effectively implement #AI, because at its core #IA creates mental models for informational systems.
22.01.2026 19:00 β π 0 π 0 π¬ 0 π 0
In a world where the major purveyors of #LLMs are going as big as possible, do the historical patterns of technological adoption encourage us, instead to go small? Scott Jenson has some thoughts.
30.12.2025 17:05 β π 0 π 0 π¬ 0 π 0
This sounds very much like #InformationArchitecture to us. So are people trained in design equipped to execute the ideas of governance and architecture? What do you need to learn to take this approach?
buff.ly/he7raav
Some people think this image is a horror story, but what we see is an informational matrix that describes of common factors when you're trying to figure out how to make more or less of a thing than is in the standard recipe.
There's a reason that Babylonian math was base twelve and base sixty!
Humans remain far better at strategy and structure than AIs. The real proof of successful agent-type AI will be when they can do this kind of high-level organization effectively instead of, essentially, copying a different version or gesturing at some aggregate idea of one.
16.12.2025 17:05 β π 0 π 0 π¬ 0 π 0
"The most effective organizations master measurement by asking one question: 'What specific decisions will change based on this number?' If the answer is vague, they refuse to measure it...true mastery [of metrics] means adjusting course when measurement undermines its own purpose."
buff.ly/7WZYBkf
This is a neat idea. "Integrity" here is closer to "integration", but in doing so invites clarity of creation, purpose, and implementation.
buff.ly/gP06hN3
If a unit is reporting to the C-suite directly, it is seen as a core capability or critical function that requires strategic oversight.
But for us the real question is: why do some organizations see UX as a C-level priority that deserves this kind of attention, and others do not?
buff.ly/MLkpnio
Just because something is easy doesn't mean it's useful. This is an emerging theme from the vibe revolution.
The more we learn about the power of AI the more we rediscover the importance of quality, understanding, and planning.
Dan Klyn uses an LLM to create "messy" ontologies from lists of words and invites you to do so with your own lists.
"Bask in the flatness" may be one of the best ways to think about the state of our information world today that we've ever heard.
buff.ly/I1Mc6Ek
"Delivery is about output...Execution is about system design: how your organization actually turns intent into outcomes. It's the invisible architecture that determines whether you're learning or just busy."
buff.ly/u1mBPCj
This essay covers Joe's thoughts on making "sh*tty" models in order to move understanding forward. We love this approach, but it's important that everyone involved know the models made quickly and incompletely are an intermediate step to completing a task, not a final product.
20.11.2025 17:03 β π 0 π 0 π¬ 0 π 0Is the fundamental structure of LLMs that of a #foxorhedgehog? A store of human knowledge loosely organized by statistical pattern matching, less accurate than associative, quickly making connections that are interesting as a tool for discovery and inspiration.
That sounds pretty foxy to us!
A really good high level description of how small AI projects can augment workflows.
People who use AI the most return, again and again, to the importance of deeply understanding what the work is in the first place.
Great thoughts from Heather on #taxonomytuesday about the problem with using the word "term" when building taxonomies.
There is a key idea here there is the NAMING of a thing (labels) and then there is the STRUCTURE of the thing (concepts). They are distinct.
Great Stuff!
A really great discussion of how to consider durable knowledge in technical projects by Indi Young.
We're glad designers are having this conversation more often, because it will make software projects better in general.
buff.ly/iSJZBUF
Archetype modeling ensures that the information structure at the core of a digital place is optimized to help users get things done. Learn to anticipate user needs in your digital places!
Starting November 13th.
More info: buff.ly/tsAjNar
#DesignNurture #TUGWorkshop
The key thing to remember about "vibe" anything is that LLMs are not procedural turing machines. They are naive, chipper magpies trying to find the things that will make you happy, for better or worse. Working with them in that mindset is critical.
04.11.2025 19:02 β π 0 π 0 π¬ 0 π 0
Archetype modeling ensures that the information structure at the core of a digital place is optimized to help users get things done. Learn to anticipate user needs for your digital places. Starting November 13th. More info: buff.ly/tsAjNar
#DesignNurture #TUGWorkshop
Taiwan has been at the forefront of using social media tools to strengthen their democracy, so much so that Audrey Tang was interviewed on the "Your Undivided Attention" podcast. This article is practical and hopeful in its ideas for using AI to improve democratic outcomes as well.
28.10.2025 16:08 β π 0 π 0 π¬ 0 π 0
We're not sure that what Jorge is describing is "conservative" as much as "average to its very bones", but his key point stands: AI is not good for finding less well lit corners or surprising results. It literally can't, and we have to keep that in consideration as we use it.
buff.ly/V5p5FFv
