- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
What is the right abstraction for context-centric work?
Posted on 7/20/26 at 2:47 pm
Posted on 7/20/26 at 2:47 pm
I'm exploring an idea that has been bouncing around in my head for years, and I'm curious whether anyone has encountered similar thinking, existing research, or products.
ChatGPT sums up what I've been trying to get at:
I realize that several desktop environments have tried to solve a similar problem by using multiple desktops or things like KDE Activities, and I use these tools often, but those are too limited and I'm not trying to build a whole new DE anyway. I'm more trying to get my head around a concept that I've been circling for developing business apps that don't inevitably end up being another menu/screens/lists/details CRUD app.
ChatGPT sums up what I've been trying to get at:
quote:
For decades, most software has been organized around artifacts:
* Files
* Folders
* Documents
* Windows
* Applications
* Pages
* Records
Even modern productivity software still tends to revolve around those concepts.
But when I think about how I actually work, I don't think in terms of files or apps. I think in terms of *what I'm trying to accomplish*. A customer quote, a manufacturing job, a software project, troubleshooting a network, planning a vacation—each of these pulls together files, emails, chats, notes, browser tabs, people, timelines, tasks, and other resources.
In other words, I seem to operate within some kind of **active slice of reality**.
I've used words like *context*, *workspace*, *thread*, *runtime*, *scope*, and *situation*, but none of them feels quite right.
One question I'm wrestling with is whether this "thing" should be considered a first-class object.
Instead of saying:
> Open this file.
or
> Launch this application.
what if the primary action became something like:
> Enter this _____.
The system would then assemble everything relevant:
* Documents
* Communications
* Applications
* AI memory
* People
* History
* Physical assets
* Dependencies
* Current state
Another question is whether this object is even something that should be *stored*. Maybe it's better thought of as a dynamic projection or query over a graph of relationships, while the user's visual arrangement of that information is stored separately.
Finally, I'm wondering whether an infinite canvas (or several coordinated canvases) is the right UI for interacting with such a thing, or whether the canvas is merely one possible representation of it.
I'm less interested in "how would you build this?" and more interested in the conceptual model.
Some questions I'd love to hear thoughts on:
* Has this idea already been explored under another name in HCI, knowledge management, philosophy, or systems design?
* Is there a better word than *context* or *workspace* for an active, evolving slice of reality?
* Should such a thing be persistent, computed on demand, or some hybrid?
* Are there existing products that truly organize computing around this concept rather than around files, documents, or applications?
* If you were designing computing from scratch today—with AI, cloud services, and graph databases available—what would you choose as the primary unit of interaction?
I'm interested in both academic references and practical software examples. Even if you think this is a dead end, I'd like to understand why.
I realize that several desktop environments have tried to solve a similar problem by using multiple desktops or things like KDE Activities, and I use these tools often, but those are too limited and I'm not trying to build a whole new DE anyway. I'm more trying to get my head around a concept that I've been circling for developing business apps that don't inevitably end up being another menu/screens/lists/details CRUD app.
3
Posted on 7/21/26 at 9:55 am to Korkstand
The conceptual leap that I don't see stated explicitly in your write-up is the primary object isn't a document or an application. It isn't even the situation itself. It's the user's current engagement with a situation. That engagement is a live process. It has memory, active AI assistance, running services, open conversations, changing priorities, and evolving relationships. It can be paused, resumed, branched, merged, or handed off to someone else. In distributed systems terms, it's almost like an execution context rather than a static record. If that framing resonates, then you're not just proposing a new way to organize software, you may be describing a successor to the desktop metaphor itself, a shift from objects on a desktop to living engagements with parts of reality. That's a direction that pieces of HCI, knowledge graphs, collaborative software, and AI have all pointed toward, but none has fully realized as the central organizing principle.
Posted on 7/21/26 at 10:44 am to Korkstand
This is good stuff but it relies on the idea that software should be written for humans at all outside of media and communications. When the human becomes the bottleneck as you describe, the simple solution is to remove the human.
Posted on 7/21/26 at 11:45 am to Breauxsif
quote:
The conceptual leap that I don't see stated explicitly in your write-up is the primary object isn't a document or an application. It isn't even the situation itself. It's the user's current engagement with a situation. That engagement is a live process. It has memory, active AI assistance, running services, open conversations, changing priorities, and evolving relationships. It can be paused, resumed, branched, merged, or handed off to someone else. In distributed systems terms, it's almost like an execution context rather than a static record. If that framing resonates, then you're not just proposing a new way to organize software, you may be describing a successor to the desktop metaphor itself, a shift from objects on a desktop to living engagements with parts of reality. That's a direction that pieces of HCI, knowledge graphs, collaborative software, and AI have all pointed toward, but none has fully realized as the central organizing principle.
Yep, that's what I'm trying to get a handle on.
Posted on 7/21/26 at 12:19 pm to Dallaswho
quote:Yep, "remove the human" is the driving force behind pretty much every (work-related) technology ever developed, whether it's "high tech" hardware and software all the way down to low tech simple machines like a wheel or lever.
This is good stuff but it relies on the idea that software should be written for humans at all outside of media and communications. When the human becomes the bottleneck as you describe, the simple solution is to remove the human.
Maybe the day that AI does everything is closer than I think, but as of now humans still find plenty of reasons to launch a particular app or open a particular file themselves. Perhaps what I am really targeting is simply a "human-in-the-loop" interface or interaction model. I think most of the software I'm working on lately can be reduced to a notification and decision-making platform. As the human, we are basically just part of the I/O stream... we receive information inputs from the computer, make a decision, then give our output back to the machine. I guess I'm just trying to streamline that role. When I switch mental contexts to work on a new decision, I don't also want to have to flip multiple sub-contexts in the interface to prepare for that.
Lots of folks here are making awesome Home Assistant dashboards. Some, possibly even I myself, would argue that the ultimate goal of HA is for it to just disappear and run everything itself, but I think having a monitor and control surface will be useful well into the future. Imagine an HA dashboard that knows what you want to see/do and it presents you with exactly what you need in a logical and intuitive layout. I guess if AI soon gets to the point where such interfaces can be generated dynamically and instantly, complete with knobs and buttons, then my whole idea becomes outmoded.
Posted on 7/21/26 at 1:11 pm to Korkstand
I’m on the side where there should be no dashboard. I have home assistant but haven’t opened it more than 2-3 times this year to work on assist. I’ll say voice commands or I have my own PWA that is much lighter and faster that I’ll use to open the garage or turn on the hot tub or anything where I’m not near a smart speaker. I work on the PWA a lot but that’s a hobby thing. If I want an automation, I navigate to /srv/ha/ and type “cline” and tell it to make it.
I guess next step would be to give the smart speaker agent full agentic coding loop abilities but I’m not that interested in doing that.
I guess next step would be to give the smart speaker agent full agentic coding loop abilities but I’m not that interested in doing that.
This post was edited on 7/21/26 at 1:18 pm
Posted on 7/21/26 at 1:54 pm to Dallaswho
quote:Yeah I think that's the goal and I think for the vast majority of things we are there.
I’m on the side where there should be no dashboard.
quote:There are still things that require our input (usually a decision/choice) and while voice commands are nice there are many cases where controls can (and IMO should) be more ergonomic than speech. Not that everything that is left needs a "dashboard" but they are the most versatile option I can think of for most things as they handle both input and output.
I’ll say voice commands or I have my own PWA that is much lighter and faster that I’ll use to open the garage or turn on the hot tub or anything where I’m not near a smart speaker.
quote:Yeah, for well-defined tasks/issues that we can articulate to a coding agent, that is of course the way to go these days.
If I want an automation, I navigate to /srv/ha/ and type “cline” and tell it to make it.
I'm thinking more about work though. There are more and more roles that we can fill with AI agents, and we have tools a level above them to orchestrate. I don't think we can, would, or should delegate *all* roles to AI, though. So some roles will always be filled by humans, even if it's only in a human-in-the-loop, guidance, or ownership capacity. So I guess what I'm trying to figure out is how can we best orchestrate a team consisting of humans and AI agents (along with other software tools). We can't use MCP or other protocols to pass messages like AI agents can (at least not yet). We have to use our senses, and while speech works in some cases it is linear/serial, intrusive, and inefficient at delivering dense information.
Posted on 7/21/26 at 2:16 pm to Korkstand
Ya in that case, I don’t think you can break the standard of having 3-4 screens or one very large one in front of your face. A lot of people have written books or viral posts on how they accomplished great things by doing very little or how they got smart listening to pod casts but those are either exceptions or even more likely people benefiting from the efforts of others or just lying for attention.
All that aside, I think people value fantasy over utility and will happily invest in an interface that gives them less work and more freedom.
All that aside, I think people value fantasy over utility and will happily invest in an interface that gives them less work and more freedom.
This post was edited on 7/21/26 at 2:23 pm
Posted on 7/21/26 at 2:18 pm to Korkstand
I'll summarize my current position and understanding like this:
As automation increases, the work that's left for humans is disproportionately about uncertainty/ambiguity, judgment, and ownership (of decisions not things). The interface should optimize for those tasks, and as of now I don't know how that should look or work, or how to even think and talk about it.
I guess I should do some reading on HCI theory.
As automation increases, the work that's left for humans is disproportionately about uncertainty/ambiguity, judgment, and ownership (of decisions not things). The interface should optimize for those tasks, and as of now I don't know how that should look or work, or how to even think and talk about it.
I guess I should do some reading on HCI theory.
Posted on 7/21/26 at 3:37 pm to Korkstand
Ya I still don’t think automation changes human I/o. If it’s for hobby/leasure, you want as frictionless and light as possible. Never seeing it is optimal.
If it’s for business, you would ideally want 100% of your effort and attention while taking in all information possible and continuously learning and improving. Big data surface, lots of information in front of you.
I spent a decade working normal hours in the office with all information and tools possible but overnights and all weekend I would rely on others to relay information to me because our business did not stop. In those cases, same as when in office and dealing with field employees, I would use iMessage/SMS or email if I wanted a record of the transaction and a voice call if I did not want a verbose record of said transaction.
I’m not sure that kind of liability filtering can be replaced by machines but also our obsession with liability is why western society is getting rapidly leapfrogged.
If it’s for business, you would ideally want 100% of your effort and attention while taking in all information possible and continuously learning and improving. Big data surface, lots of information in front of you.
I spent a decade working normal hours in the office with all information and tools possible but overnights and all weekend I would rely on others to relay information to me because our business did not stop. In those cases, same as when in office and dealing with field employees, I would use iMessage/SMS or email if I wanted a record of the transaction and a voice call if I did not want a verbose record of said transaction.
I’m not sure that kind of liability filtering can be replaced by machines but also our obsession with liability is why western society is getting rapidly leapfrogged.
This post was edited on 7/21/26 at 3:41 pm
Posted on 7/22/26 at 8:26 am to Korkstand
And just like that GitHub launches Copilot Canvases that let you collaborate with agents via a canvas. I really think these kinds of shared visual workspaces will be the way forward. Then once we incorporate gestures it could be close to a VR experience without the headgear. The sci-fi movie big interactive smart white(transparent)board UI/UX might not be too far off.
It might not be exactly like Minority Report or Iron Man, but it's looking more and more to me that we need a richer and higher-bandwidth way to work with AI, and collaborative visual workspaces look like a good way to achieve that.
Maybe now I'm just looking for the right vocabulary to discuss the development of the class of software that will be built to take full advantage of this interaction model?
It might not be exactly like Minority Report or Iron Man, but it's looking more and more to me that we need a richer and higher-bandwidth way to work with AI, and collaborative visual workspaces look like a good way to achieve that.
Maybe now I'm just looking for the right vocabulary to discuss the development of the class of software that will be built to take full advantage of this interaction model?
Posted on 7/22/26 at 9:16 am to Dallaswho
quote:
I’m not sure that kind of liability filtering can be replaced by machines but also our obsession with liability is why western society is getting rapidly leapfrogged
The most underreported issue in tech today, imo.
No one wants to just do things, everyone wants to cover their arse, because that's what we've incentivized: saying why you can't or shouldn't do something.
Page 1 of 1
Popular
Back to top
Follow TigerDroppings for LSU Football News