Published
Weekend Reading — Happy π day!
Mina "It's PiDay!"
Tech Stuff
How to Do Code Reviews in the Agentic Era I've been working on open-source for over 25 years, I suspect that's why I accept The LLM as just another project contributor:
My “unpopular” take: I don’t really care if a human or an agent wrote the code. In open source, contributions are zero-trust. Whether the code was written by a seasoned FAANG engineer or a high schooler in Sri Lanka, it shouldn't really matter. So why should we care if it was written by AI?
cosmiciron/vmprint JavaScript library to produce beautifully typeset PDF documents. Does not require browser or native app, so can run in your browser, Node, anywhere.
PageAgent A GUI agent that lives inside your web app.
What OpenClaw Reveals About the Next Phase of AI Agents "OpenClaw nailed three things that previous agent projects missed: proximity, creativity, and extensibility." Being proactive made all the difference to me and elevated OpenClaw to a level above Claude Code. I'm using Sonnet 4.6 with both, so the difference wasn't their brains power, but having OpenClaw prompt me, which led to many new interesting uses. 💡
Mud Simple and effective markdown viewer that can handle GitHub-flavored Markdown, flip between Markdown view and raw text view, and auto-reload the Markdown file, so you can edit in one app and live view the edits with Mud. (h/t Joseph Pearson)
Klaus Another OpenClaw hosted for your benefit, and this one includes AgentMail, Apollo, Hunter.io, Google Workspace, browser automation, and more. $19/month sounds like a reasonable price.
Essential Claude Code Skills and Commands Some tips for making the best of Claude Code's built-in skills and commands. I like that I can just ask it to simplify:
What makes it effective is that it spawns three parallel review agents, each looking at the changes from a different angle. You get a mini code review without leaving your terminal.
/simplifyYou can optionally pass a focus area:
/simplify memory efficiency
/simplify error handling
/simplify reduce duplication
Agent Safehouse Move fast, break nothing. YOLO mode your AI agent.
Nono Another great choice for a project name — kernel-enforced sandboxing for macOS/Linux.
Nah Continuing the trend of small apps that have "not right now" in their name, nah classifies every tool call by what it actually does using contextual rules that run in milliseconds.
AgentMail Raises $6M So AI Agents Can Finally Nag You About Unread Emails Too AI now gets its own inbox, spam folder, and all the email drama humans love to hate:
Office workers everywhere are reportedly thrilled about this development. "Just what I needed," said Sarah Jenkins, a project manager at a tech firm. "Now instead of just getting emails from my boss, coworkers, and that one client who writes in ALL CAPS, I can also get cc'd on emails between the marketing AI and the sales AI arguing about lead scoring algorithms. My inbox is going to be so much more... meaningful?"
CodeSpeak You write the spec.md and CodeSpeak generates the code for you. You may get correct code, or get not correct code, depending on how the LLM is feeling that day, how well you articulated your desires, weather you're paying the LLM a living wage, the cycle of the moon …
Eye for Design
hit-area A collection of Tailwind CSS utility classes for expanding the hit area of interactive elements. Make your app more accessible.
TUI Studio Everything you need to design TUIs like a pro. Exports to Ink, BubbleTea, Blessed, Textual, OpenTUI, Tview — if these names mean anything to you, you may be the target audience of this app.
Peoples
AI productivity gains are 10%, not 10x My experience as well:
A ~10% gain is consistent with what we’re hearing from engineering leaders more broadly: most organizations are landing in the 8–12% range. It is a real improvement, but it’s a long way from the 2–3x gains many executives and boards have come to expect. AI is moving the needle, but leaders may need to reset expectations internally.
Dare Obasanjo Related:
A study 1,488 workers who use AI tools showed some interesting results
Partícipants saw increased mental fatigue especially after using 3+ AI tools and having to oversee multiple AI agents. However they saw reduced burnout thanks to automating repetitive tasks.
I’ve argued being a manager is the best preparation for being a worker in a world of supervising AI agents instead of doing the work by hand. Now backed up with research.
One documentary-maker who’s won Emmys, he messaged me and he was like, ‘I’m being handed a shovel and told to dig my own grave,’ and that’s exactly how everyone thinks about it. ... It felt that someone put it well on Slack when they said it was like they were living in a fishbowl waiting for their human masters to drop in food, and only the ones who were fast enough to swim to the top could eat.
Business Side
Is the Atlassian Ecosystem Starting to Crack?
Atlassian’s entire economic model makes money from managing Jira complexity. Not just product complexity, but organizational complexity. The ecosystem works only because it’s genuinely and maybe even purposefully messy.
The ‘AI-Washing’ of Job Cuts Is Corrosive and Confusing FYI in a survey of 1,000 hiring managers: 59% say they stress AI's role in layoffs or hiring freezes "because it plays better"; only 9% say AI has fully replaced roles.
How My Life Got 100x Better When I Stopped Thinking About Google I'm working on a Google Search Console for AI. Does that mean I think about Google too much, or does that mean I don't give a hoot?
At some point during this process, something shifted.I stopped caring.Start24 was getting traffic through other channels — paid ads, direct visits, email, social. The site was doing fine. Not because of Google. Despite Google.And once that clicked, the weight lifted.
Machine Intelligence
Tropes "A useful tool for spotting the writing habits that make your text sound like it was written by an AI." — Claude
AI And The Ship of Theseus Slopforks: what happens when a library gets rewritten with AI?
There are huge consequences to this. When the cost of generating code goes down that much, and we can re-implement it from test suites alone, what does that mean for the future of software? Will we see a lot of software re-emerging under more permissive licenses? Will we see a lot of proprietary software re-emerging as open source? Will we see a lot of software re-emerging as proprietary?
Software Bonkers What if the biggest casualty of AI was Intuit? Would you be heartbroken?
Most miraculously: For the first time in my life, I have a dashboard that gives me a true, holistic view, of everything financial happening in my life and business. I’ve named my glorious contraption TaxBot2000. It is astounding. Let me repeat: I built this in five days.
It’s not perfect, and it’s not done, but it is already better than what I had been using for the last decade.
How we Rewrote 130K Lines from React to Svelte in Two Weeks 🤔
Instead of writing a migration proposal and spending weeks on planning, Markus built 60% of the replacement and showed it to the team. This is a key lesson in the age of coding agents: If you're trying to convince a team to do a major rewrite or feature implementation, a working demo will do more than any document.
A year ago this would be terrible advice. You'd risk weeks on something that gets scrapped. But now that building a working proof of concept with agents costs less than writing the proposal, the calculus has changed.
Pre Founder Mode Pick the one number that matters most for your startup and pre keeps you focused on it every week. I absolutely love this idea, just not sure about the product. Can't you add this as a skill to whatever AI you're already using every day? Wouldn't that get you more mileage? Why use (and pay for) a separate app? Am I foundering wrong?
feliks "It's free real estate compute"
Insecurity
When I started in security, one of the prevailing attitudes was "The weakest link in the chain will always be the human."
I would like to thank every LLM provider and startup for changing this paradigm by introducing a much weaker link in the chain.
AI Incident Database (via Sue is Writing Solarpunk)
sjvn "Easy-peasy!"
Everything Else
A giant cat and a Back to the Future reunion: photos of the day (h/t Mike Bieser)
You celebrate Pi Day. I use ISO 8601. We are not the same.
I just read a sentence that referred to “the decades since” 2005 and I feel personally attacked.
A hoodie for your nervous system New fashion trend — clothing designed for people with "busy brains":
The Stim Hoodie manages all three by incorporating unique design features into the sweatshirt that help those with autism, ADHD, and anxiety to feel calmer, more focused, and in control, per The Yorkshire Evening Post.
I was helping a younger colleague with a presentation today and said “Put a carriage return there” and with all honesty they responded with “I have no idea what that means”
I've been getting into writing a lot of short fiction.
I call them 'to do lists'
Yawn just made the only nightlight with a personality crisis This nightlight doesn't look excited to see me 😿
the three body problem is the unsolved scientific problem of what to do when you are in a group of three friends walking on a narrow sidewalk and have the unspoken struggle for who goes where in the line order and grouping
Re-reading "Neuromancer" and I'm struck how the protagonist wanting to get out of debt by selling 3MB of RAM went from being laughably dated to seeming actually plausible within recent months
Girl Scout troop sets up shop at weed dispensary. Cookies are in high demand. No comment.
Most economists agree that by the time you're 50, you should be living in a book-filled cottage at the edge of the forest and solving minor mysteries in your village with the aid of a curious ghost cat.
I have received A Lot Of Feedback (ALOF™) in the last week. The hardest and most insidious bits echo “Why does Open Source Thing need money?” which rhymes with “Why do you—a software developer working on Open Source Thing—need money?”
Trying not to read too much into it 😬
Sending a perfectly-written email? That's so low-end. First they came for our em-dash and — what would you say about that? — then they thought we don't make enough typos:
I come bearing great news for my kind of people (horrible typists):Typos are the new status symbol. Garbled spelling, a missed space, improper capitalization — those are all the new and best ways to signal to others that you are powerful and elite.
I think the reason cats are my favorite domestic animal is that the main thing going on in their brains is "I wonder if...?"
I mean, just because it usually leads to "I wonder if this object would look funny if it fell and broke" doesn't change the fact that they obviously have a great inner life.
went into full blown panic response mode earlier - one of our co-workers just started to send a couple of us really long strings of random characters on slack. no response was provided when we asked what was up - we assumed they were having some sort of medical emergency.
anyway eventually I was able to get them on the phone, which is where they informed me they had a cat that will often wander across the keyboard.
phew.
Book Corners A community-driven platform to discover, map, and contribute little free libraries around the world.
The notion of a broken clock being sometimes right is based on a gross misunderstanding of what information is.
A clock that always shows the same time is never right, even in the moments of the day when the time happens to be what it shows, because you don't gain any information about what time it is by looking at the clock.
This reasoning also applies to chatbots. If you can't tell whether what you have been given is useful information unless you alreay know the information, then you haven't been given useful information.
The Cynical, Gullible American Man The trouble with believing anything and nothing at the same time:
We have a data economy that thrives on selling products we don’t need for problems we don’t have, and a public that falls for these ploys—even as we think ourselves much too clever to be fooled.
