LLMs are currently kind of slow. OpenAI's new gpt-oss 120b is a pretty capable model in the benchmarks (maybe ~1 year behind the state of the art) If you run it on Cerebras you can glimpse the speed improvements that are coming.

Dev tooling often behaves differently if you set CI=true We're going to need AI=true so that dev tools can behave differently when called by an AI agent: - Non-interactive (like CI=true) - Less logging to preserve context - Tell the agent what commands it could run next based on what happened

When you use claude code a lot, it feels obvious that coding with agents on mobile makes sense. Seems likely >20% of 'coding' will happen on mobile within 5 years (presumably from ~0% today!)

If you have a social media app where the DMs are important but you want to stop being exposed to the feed. e.g. LinkedIn or Twitter You can use beeper.com which aggregates chats across multiple platforms. Then block/uninstall the original app.

Was explaining agentic coding to my mum yesterday and found myself saying: "My job used to be typey typey, and now it's chatty chatty" Kind of sums it up nicely.

With Claude Code you can bring it up to speed on a PR by doing this incantation: git fetch origin refs/heads/main:refs/remotes/origin/main && git diff "$(git merge-base --fork-point origin/main HEAD)" Which calculates the PR diff to main, like you'd see on Github.

Freshened up limitphone.com (again!) ✅ Visual refresh ✅ Landing pages for parents, gambling and porn ✅ New main video + videos for parents, gambling and porn. ✅ Positioning v5!

Using Claude Code on the Max plan feels a bit like fast-forwarding 2 years in the software engineering industry. It's equal parts exciting and scary.

LLMs do markdown code blocks way better if you ask them to put it in quadruple backticks.

Something I've learned about myself in the past few years is to trust my 'spidy senses' on complex problems. They turn out to be right 95%+ of the time but come with 0 rational explanation I can pass on to others. Usually takes a couple days for my brain to catch up.

I'm internally conflicted about using AI tools to code. I want it to be good! But I keep trying claude code in good faith but remain pretty unconvinced. I'm slowly diverging from quite a few of my friends and colleagues who seem are all increasing vibing.

I've recently been working on my computer ALOT and I've been pleasantly surprised how much better my body feels doing it with a split keyboard. Still better to probably spend less hours on my computer though!

llm-ui.com hits 1k stars! No matter how you're counting.

Limit now has a safe search option. Forces safe search on in Google and other search engines. Search engines without safe search are blocked.

llm-ui.com got picked up by daily.dev and doubled in stars in 2 days. Neat! app.daily.dev/posts/react-...

Still a bit sad Slack has won 'company messaging' market. I feel like it's designed for maximum engagement. It requires a lot of disapline from me / the people I work with to be productive.

There's no prebuilt of gambling apps,, and the gambling apps are different in every country. I had to use some smart techniques to build a comprehensive list of ~1500 apps. The brute force way I did it would have taken months for humans. But cost ~$30 in GPT 4o-mini.

Limit now has a Gambling Blocklist. Blocks ~1500 apps and 0.5M websites.

Been using crappy, but functional, dishwasher that came with our house for 4 years. Decided to replace it, new one is arriving today! Get up this morning and the old one is shorting the electrics (never happened before). It's making me so uneasy that I can't explain it.

git commit -m "$(claude -p 'Print a succinct commit message from my diff' "$(git diff --staged)")" This frees up at least 50% of my brain's cycles...

Limit Browser can load uBlock origin filter lists. It checks if a site is blocked in ~5ms, giving the reason if it's blocked. Performance is important because a nsfw filterlist is a 10MB .txt file!

Limit now has adult content blocking!

You can now customize how other browsers are blocked by Limit. All other Browsers were previously unconditionally blocked, but now you can customize them like any other blocklist 🔥. Shout out to that one insurance app that always opens Chrome no matter what 🙃.

🚀 Big Limit Browser update is out now: ✅ New Logo ✅ Consistent branding (much less "Firefox" mentions) ✅ New "Blocked by Limit" screen with the reason the page was blocked ✅ Blocking is ⚡ fast. This lays the foundation for large blocklists which is coming next week.

Yesterday some friends and I were comparing screen time. Mine was so low I was actually pretty surprised. Limit really works for me.

LLMs have not moved the needle in my day-to-day coding, but it's upped my game in areas like: design, video editing, seo etc. I think they help me with breadth more than depth.

Discovered "Mise", a package manager that lets you have multiple versions of a package on your machine. mise.jdx.dev/walkthrough.... It's like nvm, but you can install nvim, go etc. You can setup mise.toml in your project to override versions of tools when you cd in.

Started building adult content blocking into Limit Phone. It's easily 10x harder than blocking social media feeds. The feeds are in <200 apps and domains. Porn is so much more widespread and less centralized.

Finally found a use for chatgpt 'deep research' - building block lists for Limit Phone. "Find me all the domains for these search engines" worked pretty well.

screen.studio`s effects are easy to replicate in finalcut pro limitphone.com/demo - Use Chatgpt to help - Buy $40 plugin to do the zoom and panning - Waste 3 hours fighting final cut pro - Come full circle: and decided screen.studio is actually awesome!

limitphone.com/demo now has a proper video player (planning to put this video on the homepage). Now I just have to figure out why creating a youtube-ish player is so bloody difficult in 2025!

Huge Limit Phone redesign. Nearly every screen has been updated. Here are the highlights: ✅ Timer countdowns when unblocking ✅ New delay picker ✅ Light mode improvements ✅ Performance fixes Full demo: limitphone.com/demo Really happy with how it's coming together!

The best protein shaker is a narrow Klean Kanteen. They're high quality, last forever and have lids that seal super-well. It has a 4.5cm opening. You stuff a 5cm shaker ball in it - which won't fall out (bonus!) You then need to find a funnel. Jam funnels work well!

New Limit Phone Logo. Not perfect, but it beats the old logo which was literally from lucide icons! New fonts as well. I think it came out pretty well. Time to get a t-shirt! limitphone.com

'Vibe coding' generates a lot of code you've not really looked at. Sometimes, debugging that code wipes out all the time you saved x5 🙃

Just got a physical copy of my favorite startup book of the past couple years: Traction. It's a book for engineers which covers 20 marketing channels (by the founder of DuckDuckGo). www.amazon.co.uk/Traction-Sta...

New install method for limitphone.com just landed! Video of the install process here: limitphone.com/install-demo

Built an LLM-based app? How do you know it still works after tweaking prompts or switching models? Xata Engineer, @richardgill.org built a TypeScript Eval for the open-source Xata Agent using the @vercel.com AI SDK + Vitest. Here's how he did it 👇 xata.io/blog/llm-eva...

Just shipped a new version of Limit Phone which allows users to customize their website blocks 🌐

A YouTuber made an excellent video showing off Limit Phone. If you want to see what I've been building in my spare time - check it out! www.youtube.com/watch?v=LeW0...

I've been steadily building limitphone.com in my spare time. This morning when I woke up I'd gotten my first paying customer!