
Helping people build with AI and nocode @motioninproduct 🇨🇦 🇬🇧 🇦🇺 #ux
Sydney, New South Wales
Joined March 2009
- Tweets8.3K
- Following2.3K
- Followers916
- Likes12K
Sometimes you accidentally create a tool worth sharing when you prototype an idea.
What if you could just drag and drop files and have it running in GitHub pages. I created a tool inspired by Vercel Drops to explore this workflow.
mycodedrops.xyz/
These posters from the DnB scene in Cardiff between 2005-2017 is some elite design work by @danmizziharris aperture.danmh.xyz/
Michael Le retweeted
a few people asked me how i timed the raccoon animation on my portfolio!
i basically asked claude to generate me a red line to visualized its path, and used @joshpuckett's dialkit to tinker around (adjusting the slope, peaks, etc.)
Working with a stricter token budget means you need to focus more. Rather than spinning 3 worktrees and working on 2 ideas at once. It forces you to focus because you'll never finish anything otherwise
AI makes it easier to build custom tools but it doesn't mean that all of those tools will become companies.
Michael Le retweeted
In Plannotator Code Review, you can already launch review agents - Claude and Codex - they leave annotations on diffs. These are the same class reviews those harnesses provide.
There is also a code tour feature. Personal experience shows these are not as useful as just chatting with a model when trying to learn about changes
1. Do you want the ability to create custom reviews?
2. Is there heart burn if the code tour feature is removed?
On 2, the better path is deterministic semantic diffs used strategically. And more interactive "tour"/debugging like features beyond simple summaries you couldn't otherwise get from chat.
Thoughts?
Instead of tracking token usage and pull requests individually to measure AI usage, it should be a measure of a ratio of token usage per pull request that got merged towards moving a metric.
Counter measure is how many of those PRs got reverted/revisited
Managing AI token spend is the corporate version of managing your text messages when it used to cost 10 cents per SMS
Michael Le retweeted
If you want to know what giveaway LLM design is, this is it: thin colored borders, gradients, glow effects, too many different font sizes, small fonts too small, inconsistent padding and alignment (especially vertical).
Not a dunk on @zeeg he’s being transparent about it. And he’s a good AI driver and good developer in general. But using this moment to show people how obvious this is.
For those who don’t know, the HTML Canvas allows you to generate powerful and performant graphics, but it's terrible for SEO/AIO because it doesn't render real DOM text elements for readers.
Historically, we’ve always had to separate canvas elements from other DOM elements: like HTML text elements overlayed on top of a canvas background (in technical terms, these are basically sibling elements).
What this new API unlocks is it allows HTML elements to become child elements of a Canvas, blending graphics and HTML elements seamlessly.
We're going to have some crazy UIs in the near future!
Design being shared these days on X and Mobbin is so good. It’s been years since it has been like this for myself
Michael Le retweeted
The rise of AI-assisted coding (vibe, agentic, etc.) is really putting pressure on designers and I don’t think it’s being talked about that much
Let’s think about how software at tech companies used to be created, broadly:
Some people would figure out what the next thing to build should be. Designers would produce concepts and narrow to the final experience. Engineering feasibility would be assessed, then the thing would be built, tested, and shipped to users
Within this process is the expectation that Design Would Be Ready before Engineering Builds It because if not, then it gets messy, additional cycles are wasted, or engineers get sick of waiting and stop respecting designers and just ship whatever they want
I’ve worked at companies where design had a few days, a few weeks, or a few months before Engineering Builds It, and obviously more time is better but that’s usually a luxury because screens need to get designed and approved as fast as possible
But now with AI everything can be built instantly, all features can be coded by agents while humans sleep, and things that used to take a week to build now take less than a day, or less than an hour, or maybe built by the PM on a Saturday night
So when do things get designed by designers?
Well, optimally, everything that could ever be built has already been meticulously designed and thought about and iterated on and is just sitting in a Figma file waiting for the water to boil
But in reality, Good Design takes time because it takes some thinking and talking and sitting with the problem and iterating and reworking and AI reduces that time a lot but it doesn’t drop it to 0
You’re probably already thinking that the solution is to have a comprehensive Design System that AI and AI-assisted engineers can reference and pull from, but the design of an incredible new feature cannot spontaneously emerge from a library of panels and buttons and toasts and color tokens because that’s like going into the fasteners aisle at Home Depot and expecting the architectural blueprints of your dream home to manifest out of a bucket full of nails
PMs can use AI to figure out what to build next, engineers can use AI to build it, so everybody is just waiting on those Figma mockups that were due yesterday and when talking to my product design friends at big tech companies about this it’s clear that they are stresssssssed
Leaders have a few options to mitigate this:
First, they can bring designers in earlier and get them started concepting and iterating while The Idea is still being refined (the best tech companies already do this)
Second, they can push for designer adoption of AI just like engineering has been pushing it. Give designers all the tools (MJ, Veo 3, Visual Electric, ElevenLabs, Krea, and a half dozen others) and hire consultants to give real world demos of how they work and how to multiply creative output while shrinking time spent
I left the corporate tech world over a year ago, but if you still work as a product designer at a big tech company how are you doing? How have your processes changed? Are you staying afloat?
Michael Le retweeted
🎵💿Built an MCP that lets Claude talk directly to Ableton. Now you can create music with just prompts!
Here’s a demo of me creating a lush, 80s synthwave track in just two prompts. It picks the right instruments, creates melodies, and adds effects like reverb and distortion 🔊
Yesterday a simple prompt I gave Claude generated a React component.
Today I used Claude to create a whole Vue app complete with paginated queries and state management.
I was impressed by both experiences in quality of output and ease of use.
Michael Le retweeted
DeepSeek just proved the 'worthless' GPT wrapper startups are actually the ones with real moats.
A week ago, nothing was more LOW status than being a 'GPT wrapper' startup.
But I think we're learning that's DEAD wrong. Turns out they were just early to the only game that matters
While DeepSeek, Meta, Anthropic and Microsoft battle over benchmark scores, these 'wrapper' companies have been quietly building the only moat that matters: interface loyalty.
Because of how big owning the LLM is for national defense and the economy, the next breakthrough model is always 2 weeks away. DeepSeek launches today, someone else drops a better one tomorrow.
But getting millions of people to make your product part of their daily workflow - that's the real barrier to entry. ChatGPT didn't win because it had the best model. It won because it was dead simple to use. And I think it has staying power because of that.
This is why all those AI startups we dismissed are actually positioned to win. They're not competing on model performance - they're competing on being the default way humans interact with AI.
Frontier models are becoming commodities. User habits aren't. While everyone obsesses over the next architecture breakthrough, the real game is being played in the interface layer. The moat isn't in the model - it's in being the tool people reach for without thinking.
Technology advantage is temporary. Interface lock-in is forever.
Keep shipping those wrappers, my friends.
Michael Le retweeted
Any time you solve a problem with more than ten browser tabs open, you should be obligated to write up the answer so other people can find it in the future.
Michael Le retweeted
So what does the inverse look like?
Start the company with the goal to be the best for something (@linear is the best tool for software teams to plan and build). Each market can only support 1-2 winners, so if you are not going to be the best, you probably have already lost.
Replying to @shreyas
Probably a host of reasons:
- pm/teams make decisions based on a/b testing/data
- engineers and designers are not listened or contribute to the product direction
- teams are incentivized by “y metric go up” so vision arguments are not listened
- hiring those people who say “we launched x, y metric went up 2%” because quality or vision harder to interview
- multiple layers of management and hiring massive teams where any vision there might be gets diluted away
- product development process is framework driven/overly complex where mental pm energy goes to running it, meetings, not focusing on the product
- founders/execs focus on timelines and launch dates, not quality output
Essentially great work and quality is hard to measure so organizations give up on it.
What replaces is different kind of motions that supposedly do something towards greatness but whole organization mostly just a phoning in/acting product management instead of truly focusing on building great products.
At @linear basically trying to do the opposite of all these things :)












