263 karma · joined September 3, 2021
We will just repeat the same pattern, have the latest and greatest frontier models rewrite the software just like how many desktop monoliths moved to React SPAs.
The real winners are frontier labs, we pay them to write black box software and we pay them to rewrite the software.
Better results meaning it gets closer to what I had imagined in my mind or the design is subjectively better.
In my experience, Figma Agent and Figma's MCP leave something to be desired. I'm sure Figma Agent is useful for people who haven't used Claude Code or similar before, but it really requires most of your product's context to exist in Figma for it to do well.
As soon as your product's context exists out of Figma, then it starts to drift and hallucinate its own styles. Using the MCP with an AI that has access to your code provides better results, but even it will have trouble remembering what page I'm asking it to edit.
I know they're trying to address this by adding features such as guidelines and specs, along with linking your Figma files to GitHub repos, but my complaint with these features is a lot of the companies I have worked for do not use the very specific tech stack that Figma supports with these features. If it's anything like code connect, it may take years to get more broad support. Every company I have worked for is a C# Microsoft/Azure house.
This does not help with my broader frustration with Figma which is there is a lack of focus/vision. They're trying to compete on all fronts by developing all these different products/features and it doesn't really feel like they have a cohesive story for what the future or product design is. Granted, I think most of us in this industry don't either, but when am I supposed to use Figma Make vs. Figma Agent vs. Code layers?
They keep blurring the line between them when in all honesty I'm more interested in just using the AI agent I pay for to code what I need and bring it into a canvas for archiving/exploration when I need it.
1. The trackpad on it is so god awful that using any sort of multi finger gesture is a non-starter. In this case the touch screen works significantly better.
2. I wanted to draw in a whiteboarding tool. Drawing with my finger is not ideal so I tried a passive stylus. Drawing with a passive rubber-tipped stylus leaves a lot to be desired. I would much rather use an active stylus or a pen display for artists if I could.
Apple is known for the best trackpads. They have set the bar high and they have consistently met that bar for over a decade. I don't need a touch screen for gestures when their trackpad feels just as good. To my next point, there is no pen support on this screen.
Given this, I feel the touch screen is going to be seen just like the touch bar, a pretty feature that receives some good niche support, but ultimately does not align with how the majority of users use a laptop. I just hope there's an option to turn off the touch screen when they inevitably start shifting this display down to the base model.
For example, for DoorDash, I should just receive order status notifications but they ping me every few days to try out a local restaurant. They want to remind me that their app exists on my phone and I can spend money with them.
Ideally I should be able to turn off these kind of marketing notifications separate from functional notifications, but not all apps are this kind.
It makes me realize that email is a much better system for this. You have an entire archive of every email you receive, with the ability to search, filter, forward, and even set up custom rules to delete items containing certain text, for example.
Yes it's a hassle to manage an email inbox, but this kind of flexibility gives users a lot of power that phone, sms, or native notifications lack.
This isn't really a case of someone overlooking previous guidance, but a symptom of an org where platform consistency is being prioritized over tradition. This means changes now cater to the lowest common denominator (iOS). Lots of recent updates this, like the new app icons, the macOS 26 implementation of liquid glass, etc. The priority is making macOS easier to adopt for people who's first devices were iPads.
It's unfortunate. Although what I find strange are some of the very obtuse patterns they refuse to change. It took them forever to add window snapping to macOS and Finder still doesn't have a cut command.
Announcement thread: https://x.com/vladmoroz/status/2107865714812162228
Code connect was supposed to bridge this gap, allowing users to define how Figma components should generate relevant code snippets, but it only worked with React and they had some janky string based templating language for everything else. And of course, it meant someone would have to go through the effort of creating mappings for every component in your system. It's like writing a component twice, the same pitfall we were trying to avoid in the first place.
Now people will just ask their AI to pull the frames from Figma through the remote MCP and have the AI implement it that way. But to that end, why not just have designers make a branch in the codebase and have them implement the UI? This is the question many of us have been asking ourselves lately.
They continue to iterate though. They've introduced Code Layers so real codebases can render on the canvas, but in infamous Figma fashion it only works with a limited set of React codebases. Figma Make (their Lovable, Bolt.new, Claude Design like product) can also pull in existing codebases, but again it is very limited in the types of code bases it can work on.
These limitations are all so exhuasting. It's just so much easier to use Claude Code or Codex to do what I need. At this point Figma is more of a secondary tool for when I need to document a prototype or want a canvas for exploration purposes.
[1] https://help.figma.com/hc/en-us/articles/32132100833559-Guid...
The way I have made peace with it in my mind is at the end of the day, I am the expert in UI/UX. There are many product teams I've joined that have operated without a designer and you can tell (bless their souls). Component libraries, templates, articles, video courses, and etc. have all existed throughout this time so it's not like they've been operating completely blind to design and UX. I don't think AI would be different. Sure it may raise the floor a little, but in the end someone needs to evaluate the output and hold responsibility for the UI/UX.
It was an uncomfortable idea to come to terms with though, I like many other designers spent a decade getting good at Figma. Figma and design almost felt intertwined for a moment, but designers have a long history of having their craft disrupted by new technology. Decades ago designers were cutting and pasting paper, then moved to digital publishing; and in product design we've gone through software such as Photoshop, Fireworks, Sketch, and Figma, just to name a few. Design has survived all this and will continue to survive in the future.
I think the same applies to the other disciplines. Sure I could have AI spin up a backend, but I don't really have the expertise to understand whether what it is doing is good or full of security vulnerabilities.
I don't know in the future if we will create a new path for product builders, individuals who are knowledgeable in all aspects of product management, software engineering, and product design. I also don't know if teams will have as much need for individuals with the efficiency gains from AI.
I couldn't believe it when file annotations are only visible to users with design and dev seats. Like my PMs will never be able to read the annotations. I stopped using that feature entirely after that, and just stuck to pasting in FigJam sticky notes instead.
Figma's main value used to be in providing designers a canvas to iterate and explore ideas since the majority of designers did not code, but AI has completely changed that.
I fear Figma's reluctance to integrate with all the popular AI tools might actually accelerate their decline. AI provides so much value, that I would rather base my software purchasing decisions around what is compatible with my AI of choice rather than pick an AI that is compatible with Figma.
I only found out about Figma's limitation when I was trying to add the remote MCP server to GitHub Copilot Desktop and kept running into errors. Turns out they whitelisted GitHub Copilot CLI but not the Desktop app and had put a pause on enabling any more vendors. Eventually someone (not sure which side) got it working.
Kind of strange to limit edit access only to the Remote MCP when their competitors like Pen[1] and Paper[2] allow any local agent to edit.
Ideally everyone would just have the discipline to just keep what they need on their phone without trying to fix their problem with an overpriced gimmick that they will give up on after a few months.
It would be one thing if we maintained the same quality of service that TV had and had streaming as an alternative, but streaming has pretty much replaced TV. Every channel always has ads for their streaming service. It's bleak.
For example, I have a variety of USB C to USB C cables but they each have their own capabilities. Some support high transfer speeds, so I keep them around for connecting SSDs, while others are just USB 2 speeds so they are strictly for charging.
Another case is I have a very specific USB C to USB C cable that looks incredibly generic, but it is for connecting a screen tablet to a laptop because the manufacturer decided to make the port recessed, so only their cable fits. If this cable wasn't labeled as for a specific device, I would have donated it years ago.
It is some effort, for those who don't like to tinker with tech I can understand why they just shove them all into a box in the back of their closet. But I find that culling my cables at least makes me aware of what I have and don't have.
One other suggestion I have is just buying a longer cable than you think you need because eventually you'll have the need for it anyway. I have so many 3.5mm aux cables because I kept changing my desk set up and needing to buy longer and longer cables. If I had just bought the longest cable I could have at the time I would have saved myself some headache.
Another tip is adapters. Sometimes it's easier to just have cables of one type and a bunch of adapters for when you need them rather than having a ton of different cables for each possible scenario. I used to have a mix of RCA and 3.5mm cables but never had the right combination of what I needed at the time. Eventually I caved in and just bought a ton of adapters so I could standardize all the cables and use adapters as needed. Same with USB C cables.
I was surprised that the first thing I was prompted to do was to download an app and sign up for an account. The TV would not work unless I was signed in (or at the very least the option was so well hidden I didn't see one). Of course the sign up asks for PII like my name and phone number.
Vizio was bought out by Walmart a while back, with the intent of expanding Walmart's advertising empire.
It's really unfortunate that the TV market has ended up the way it has. My family still has a few CRTs going that have been working every day for decades, just turn it on and TV immediately starts playing. I miss that level of simplicity.
Screen readers allow users to use their keyboards and other assistive devices to navigate websites and apps by tabbing through elements with extra controls and shortcuts to skip to regions, interact with elements and such.
To that degree I agree that all GUIs should be keyboard accessible.
This is obviously not what the original author had in mind, if he had to tab through every item in a web app he would probably get frustrated rather quickly. What he is imagining is like TUIs where all the interactions are through keyboard shortcuts and macros. These are not popular with the majority of users of modern day software, especially if the software is something that is used infrequently. I think we take for granted how component we, people who create software, are.
What has become more popular are interfaces that expose extra controls with a single key press, where the user can type to get the result they want. So instead of memorizing the keyboard shortcuts to dozens of actions, you can press something like Crl+K or / to summon a palette where you can search for the action.
Claude's models in my experience do a better job of inferring my intent, or to say it does a better job of giving me the result I imagined in my mind. A recent example was a UI prototype I was building for a desktop application. I had asked GPT's 5.6 Sol to update the open document in the prototype to better reflect the context of the feature I was designing, and 5.6 Sol took it very literally and had just added some text to the currently open document, not what I had in mind. I tried again with Claude Opus 5 and it added a completely new tab with a complete new document that, although imperfect, much better matched my expectations.
You could say this was a prompting skill issue, but seeing how many people are prompting their AI I believe the labs are incentivized to continue to improve their ability to infer intent.
When it comes to the desktop applications though, I find Claude Desktop's output to be incredibly verbose and full of jargon. I feel like it hits me with an entire essay and the UI doesn't have enough typographic hierarchy to make it easy to scan. ChatGPT Desktop is much better in this regard, I feel the output is concise, clear, and gives me just enough info to feel in the loop without being overwhelmed. Even though I have the setting on for technical language, it feels more understandable than Claude. I also feel that ChatGPT's desktop app has a better design and much more polish.
I do not really like how bloated both applications have become though. This weird segmentation of Chat, Work, and Code all just seems like it's pushing a technical limitation onto the user. The other day I opened a document in ChatGPT and asked it to do something, then it told me it could only do it in work "mode", so it then created an entirely new conversation with a reference to the previous conversation. It wasn't a completely new area of the UI either, it just added a "Work" badge to the new conversation in the list. Feels a bit unnecessary, like couldn't you just keep it all within the same conversation?
I'm not sure why this issue is so prevalent, it's not hard to point Claude at the Wikipedia article on signs of AI writing or ask Claude to write content anyone of the average American reading level could understand.
To me it just gives off a sense of laziness, that you cared so little of your content that you did not take the time to read it yourself and edit it to effectively communicate the message you wanted to communicate. To that point, it's just not worth my time reading, otherwise my eyes glaze over trying to read between the lines of machine written language for other machines.
I previously really enjoyed the Codex app. It was easy for me to understand and find what I needed. The conversation view provided the right amount of information without feeling too concise or full of jargon and the typographic hierarchy made it easy to skim. The project based workflow made sense for my needs.
Ever since they combined all ChatGPT into one app though it has become a mess. It's trying to do too much in one. The settings menu is especially bloated, and I find the UI to be too complicated for the simple questions I'm asking outside of development work. I would rather just use the web interface for general use at this point. They've also been updating much of the UI to follow the same patterns as Claude, and I've not been impressed with the Claude desktop app.
To me, they are taking multiple steps backwards. I hope they eventually find their way again.
I'm not sure what in the training set causes the models to do this, even in my own experience I'll be specific about the feature I'm building and the model will add an unnecessary status message for no reason.
Missing small details like these makes it fall into the uncanny valley. It looks like a typical puzzle on the surface but when you try to solve it all the mistakes stick out.
With many leaders in tech claiming that AI agents have drastically reduced the time and cost of shipping features, how do such small usability problems still not get fixed after years?