Maggie Appleton
@maggieappleton.com
Design engineer playing with AI and hacky prototypes @githubnext.com
Adores digital gardening, end-user development, and embodied cognition. Makes visual essays about design, programming, and anthropology.
📍 London
🌱 maggieappleton.com
Adores digital gardening, end-user development, and embodied cognition. Makes visual essays about design, programming, and anthropology.
📍 London
🌱 maggieappleton.com
SVG and CSS. I gave it some existing examples from my blog and style files (global CSS vars, etc.). Plus some guidelines: be subtle, use soft shadows and round borders, use these colours
Agree it's not bad as a starting point. And seeing what I dislike helps me take it in a better direction.
Agree it's not bad as a starting point. And seeing what I dislike helps me take it in a better direction.
January 11, 2026 at 11:27 AM
SVG and CSS. I gave it some existing examples from my blog and style files (global CSS vars, etc.). Plus some guidelines: be subtle, use soft shadows and round borders, use these colours
Agree it's not bad as a starting point. And seeing what I dislike helps me take it in a better direction.
Agree it's not bad as a starting point. And seeing what I dislike helps me take it in a better direction.
It's OpenCode! It's delightful. With the theme catppuccin-macchiato. Also using Ghostty for my terminal and have same theme set there.
January 11, 2026 at 11:22 AM
It's OpenCode! It's delightful. With the theme catppuccin-macchiato. Also using Ghostty for my terminal and have same theme set there.
Oh I didn't know it was customisable at all! I'll take a look. Maybe models will be better at generating initial ideas with a more structured language like that, and I can work off it
January 11, 2026 at 11:20 AM
Oh I didn't know it was customisable at all! I'll take a look. Maybe models will be better at generating initial ideas with a more structured language like that, and I can work off it
My theory here is the major labs have done very little or no RLHF on visual explanations and diagrams? Seems like a blind spot. Image generation alone or text generation alone are very well optimised but the combination seems uniquely hard for current models.
January 11, 2026 at 11:15 AM
My theory here is the major labs have done very little or no RLHF on visual explanations and diagrams? Seems like a blind spot. Image generation alone or text generation alone are very well optimised but the combination seems uniquely hard for current models.
The best diagram outputs an LLM can get to are still crap by my standards. It helps speed me up a bit and give me starting points, but it's still disappointing that their visual reasoning and graphic design skills are really weak.
The layouts and visual language are always predictably boring too
The layouts and visual language are always predictably boring too
January 11, 2026 at 11:13 AM
The best diagram outputs an LLM can get to are still crap by my standards. It helps speed me up a bit and give me starting points, but it's still disappointing that their visual reasoning and graphic design skills are really weak.
The layouts and visual language are always predictably boring too
The layouts and visual language are always predictably boring too
Yeah I've used mermaid diagrams before, but my understand is they're very simple and standardised?
The final output I'm going for with is pretty refined. These are SVG diagrams for one of my essays so I need a lot of control over the visual aesthetics. And I usually add in animation and interaction
The final output I'm going for with is pretty refined. These are SVG diagrams for one of my essays so I need a lot of control over the visual aesthetics. And I usually add in animation and interaction
January 11, 2026 at 11:10 AM
Yeah I've used mermaid diagrams before, but my understand is they're very simple and standardised?
The final output I'm going for with is pretty refined. These are SVG diagrams for one of my essays so I need a lot of control over the visual aesthetics. And I usually add in animation and interaction
The final output I'm going for with is pretty refined. These are SVG diagrams for one of my essays so I need a lot of control over the visual aesthetics. And I usually add in animation and interaction
Claude's second attempt at fixing some of the problems, with a lot of additional steering from me on improving the visual details.
It's still unable to make an arrow go where it needs to, or make two elements connect with a line.
But it's a decent starting point - enough for me to cleanup by hand.
It's still unable to make an arrow go where it needs to, or make two elements connect with a line.
But it's a decent starting point - enough for me to cleanup by hand.
January 11, 2026 at 11:07 AM
Claude's second attempt at fixing some of the problems, with a lot of additional steering from me on improving the visual details.
It's still unable to make an arrow go where it needs to, or make two elements connect with a line.
But it's a decent starting point - enough for me to cleanup by hand.
It's still unable to make an arrow go where it needs to, or make two elements connect with a line.
But it's a decent starting point - enough for me to cleanup by hand.
I fully support ignoring a lot of what’s happening in agent hype town if it’s doesn’t speak to you. The hype will hype either way. Other useful and interesting work needs to be done in software outside of the hot new thing. Work that is likely being ignored/abandoned as people jump on the bandwagon
January 4, 2026 at 9:18 PM
I fully support ignoring a lot of what’s happening in agent hype town if it’s doesn’t speak to you. The hype will hype either way. Other useful and interesting work needs to be done in software outside of the hot new thing. Work that is likely being ignored/abandoned as people jump on the bandwagon
It's so hard to keep abreast of what's possible with models at the moment. I definitely suggest playing with Claude Code, Codex, & Gemini to develop your own felt sense of where we're at. Even just toy projects or personal benchmarks. They'll both surprise you and face plant. But in interesting ways
January 3, 2026 at 1:27 PM
It's so hard to keep abreast of what's possible with models at the moment. I definitely suggest playing with Claude Code, Codex, & Gemini to develop your own felt sense of where we're at. Even just toy projects or personal benchmarks. They'll both surprise you and face plant. But in interesting ways
Ha! It is near impossible to make the overton window inside the bubble seem reasonable to everyone outside it
January 3, 2026 at 11:05 AM
Ha! It is near impossible to make the overton window inside the bubble seem reasonable to everyone outside it
I hadn't encountered much of Yegge before so this was quite the onboarding
January 3, 2026 at 11:04 AM
I hadn't encountered much of Yegge before so this was quite the onboarding
Thank you ❤️
Rereading it this morning I realise it sounds like I am *very* tired 😅 which I am, but also very excited and full of the good things in life as well.
Writing it helped me unclench a bit and just start putting some thoughts down. But I will keep up the chocolate hobnob intake levels.
Rereading it this morning I realise it sounds like I am *very* tired 😅 which I am, but also very excited and full of the good things in life as well.
Writing it helped me unclench a bit and just start putting some thoughts down. But I will keep up the chocolate hobnob intake levels.
January 3, 2026 at 11:04 AM
Thank you ❤️
Rereading it this morning I realise it sounds like I am *very* tired 😅 which I am, but also very excited and full of the good things in life as well.
Writing it helped me unclench a bit and just start putting some thoughts down. But I will keep up the chocolate hobnob intake levels.
Rereading it this morning I realise it sounds like I am *very* tired 😅 which I am, but also very excited and full of the good things in life as well.
Writing it helped me unclench a bit and just start putting some thoughts down. But I will keep up the chocolate hobnob intake levels.
...and language is a poor medium for designing easing curves and aesthetic feelings. I always need to touch the code, and it's often faster to CSS than prompt.
FE requires constant human-in-the-loop checks so even if you have a dozen agents working, I am the bottleneck. Scale isn't that helpful.
FE requires constant human-in-the-loop checks so even if you have a dozen agents working, I am the bottleneck. Scale isn't that helpful.
January 3, 2026 at 9:02 AM
...and language is a poor medium for designing easing curves and aesthetic feelings. I always need to touch the code, and it's often faster to CSS than prompt.
FE requires constant human-in-the-loop checks so even if you have a dozen agents working, I am the bottleneck. Scale isn't that helpful.
FE requires constant human-in-the-loop checks so even if you have a dozen agents working, I am the bottleneck. Scale isn't that helpful.
I'm in fruitful debates with colleagues over projects like this. Namely: how much hands-on access to code do we think developers will need in the next generation of tools? How close does it need to be?
I'm in the code-must-be-close camp, at least for the next 2 years. My bias is I do front-end...
I'm in the code-must-be-close camp, at least for the next 2 years. My bias is I do front-end...
January 3, 2026 at 9:02 AM
I'm in fruitful debates with colleagues over projects like this. Namely: how much hands-on access to code do we think developers will need in the next generation of tools? How close does it need to be?
I'm in the code-must-be-close camp, at least for the next 2 years. My bias is I do front-end...
I'm in the code-must-be-close camp, at least for the next 2 years. My bias is I do front-end...
I'm not going to claim I entirely understand the workflow primitives even after a couple reads (does this diagram help? no? no)
But the gist seems to be breaking tasks down into atomic units of work, tracked in git, executed, checked, and merged through a zoo of loops and pre-defined workflows.
But the gist seems to be breaking tasks down into atomic units of work, tracked in git, executed, checked, and merged through a zoo of loops and pre-defined workflows.
January 3, 2026 at 8:53 AM
I'm not going to claim I entirely understand the workflow primitives even after a couple reads (does this diagram help? no? no)
But the gist seems to be breaking tasks down into atomic units of work, tracked in git, executed, checked, and merged through a zoo of loops and pre-defined workflows.
But the gist seems to be breaking tasks down into atomic units of work, tracked in git, executed, checked, and merged through a zoo of loops and pre-defined workflows.
There's a central orchestrator, implementation monkeys, a git merge resolver, an unsticker who troubleshoots issues, a work hustler, maintenance workers, special ops teams.
Agents with specific roles working in a cohesive flow with trad programming supports: git, daemons, issue tracking, tmux
Agents with specific roles working in a cohesive flow with trad programming supports: git, daemons, issue tracking, tmux
January 3, 2026 at 8:47 AM
There's a central orchestrator, implementation monkeys, a git merge resolver, an unsticker who troubleshoots issues, a work hustler, maintenance workers, special ops teams.
Agents with specific roles working in a cohesive flow with trad programming supports: git, daemons, issue tracking, tmux
Agents with specific roles working in a cohesive flow with trad programming supports: git, daemons, issue tracking, tmux
3. The conceptual design here is all over the shop – polecats, deacons, dogs, boot the dog, seancing, beads, epics, molecules, protomolecules, MEOW stack, convoys, patrols, refinery, workflows.
I got lost somewhere after the MEOW stack...
But these point at the shape of future agentic primitives:
I got lost somewhere after the MEOW stack...
But these point at the shape of future agentic primitives:
January 3, 2026 at 8:44 AM
3. The conceptual design here is all over the shop – polecats, deacons, dogs, boot the dog, seancing, beads, epics, molecules, protomolecules, MEOW stack, convoys, patrols, refinery, workflows.
I got lost somewhere after the MEOW stack...
But these point at the shape of future agentic primitives:
I got lost somewhere after the MEOW stack...
But these point at the shape of future agentic primitives:
2. The cost could be as much as $2k+ / month? (wild speculation - open to more accurate guesses)
At this early stage the efficiency of the system is literal shit, so the cost ≈ valuable output equation is inaccurate. But companies would pay ~near this for a high quality, low waste version of this
At this early stage the efficiency of the system is literal shit, so the cost ≈ valuable output equation is inaccurate. But companies would pay ~near this for a high quality, low waste version of this
January 3, 2026 at 8:44 AM
2. The cost could be as much as $2k+ / month? (wild speculation - open to more accurate guesses)
At this early stage the efficiency of the system is literal shit, so the cost ≈ valuable output equation is inaccurate. But companies would pay ~near this for a high quality, low waste version of this
At this early stage the efficiency of the system is literal shit, so the cost ≈ valuable output equation is inaccurate. But companies would pay ~near this for a high quality, low waste version of this
Yegge is certainly a flavour.
This so ahead of the curve I'm considering it as a piece of speculative design.
Interesting:
1. Design and planning becomes the bottleneck when you hand all production code over to agentic systems. Design is a bottleneck *even when* the interface is just a TUI...
This so ahead of the curve I'm considering it as a piece of speculative design.
Interesting:
1. Design and planning becomes the bottleneck when you hand all production code over to agentic systems. Design is a bottleneck *even when* the interface is just a TUI...
January 3, 2026 at 8:44 AM
Yegge is certainly a flavour.
This so ahead of the curve I'm considering it as a piece of speculative design.
Interesting:
1. Design and planning becomes the bottleneck when you hand all production code over to agentic systems. Design is a bottleneck *even when* the interface is just a TUI...
This so ahead of the curve I'm considering it as a piece of speculative design.
Interesting:
1. Design and planning becomes the bottleneck when you hand all production code over to agentic systems. Design is a bottleneck *even when* the interface is just a TUI...
Ha, yes exactly this! I love this meme.
December 30, 2025 at 6:31 PM
Ha, yes exactly this! I love this meme.
I’m very prompt happy when it comes to building small personal tools for not-too-complex workflows.
But for big, complex, and substantial workflows, I’m going to pay SaaS rates in exchange for someone thinking through all those design decisions for me. Eg. TaxScouts, Lightroom, VS Code
But for big, complex, and substantial workflows, I’m going to pay SaaS rates in exchange for someone thinking through all those design decisions for me. Eg. TaxScouts, Lightroom, VS Code
December 30, 2025 at 5:48 PM
I’m very prompt happy when it comes to building small personal tools for not-too-complex workflows.
But for big, complex, and substantial workflows, I’m going to pay SaaS rates in exchange for someone thinking through all those design decisions for me. Eg. TaxScouts, Lightroom, VS Code
But for big, complex, and substantial workflows, I’m going to pay SaaS rates in exchange for someone thinking through all those design decisions for me. Eg. TaxScouts, Lightroom, VS Code