Sohil Pandya

Pi was built when there were already agent harnesses around. Here’s why Mario Zechner(@badlogicgames), found them suboptimal and built Pi, a minimalist self-modifying agent: #1 - Mario initially was a believer in Claude Code: "I was a believer in Claude code because they were the first that packaged agentic search up in a really compelling package. And at the time that fit my workflow really well. Everything around the LLM was kind of nice and tidy and easy to understand. I was super happy. I was proselytising Claude code." #2 - Reverse engineering Claude Code highlighted the degradation that Mario felt as a user: "I personally like simple tools that are stable and that I can rely on. Even if they have non-deterministic parts, all the deterministic parts should be as stable as possible. That was just not the experience with Claude Code around summer 2025. They would take away your control of the context. They would inject stuff behind your back, which is bad. Then, your workflows stopped working because there's now a system reminder that you don't even see in the UI that would modify the behaviour of the model. They would also do this to the system prompt. I built a little service where I can track the progression or evolution of the system, prompt and tool definitions and, with every release, it was messing with stuff. That just messed with my workflows and I don't appreciate that." #3 - PI was built with an appreciation for simple and reliable tools: "If I commit to a development tool, I want it to be a stable, reliable thing like a hammer. I don't want my hammer to break a different spot every day. That's terrible. We need somebody who goes the full velocity kind of way. But I don't want to work with a tool like that."

Good engineers say 'no', a lot. Mario Zechner(@badlogicgames) and Armin Ronacher(@mitsuhiko) - creators of Pi and Flask - on complexity, agents, and why battle scars matter: Mario - saying ‘no’ keeps complexity down: “A good engineer is an engineer that says ‘no’ a lot, and ‘I don't need this’ a lot, because that keeps complexity down. If you're using agents, the exact opposite happens. You say, yes, I want this and I want this, because I don't have to type it myself. I don’t have to think about it. I just give the little machine a prompt and it'll spit out something that kind of looks like the thing I wanted. Good enough. And that's where all the problems start.” Armin - human-gained battle scars are foundational for teams to say no: “Good engineering is all about knowing the trade-offs that you have to make. I think Cal Henderson once said ‘you do the dumbest solution first until it doesn't work anymore', because there's so much to do that if you actually do the correct solution, you're creating the kind of complexity that kills you at scale. The engineer learns that — but if you don't have that battle scar, it's very hard for you to argue correctly, because it is this learning process that gives you the authority to then convince other engineers in the engineering org that you should be doing it this way.”

GitHub’s reliability is less than one nine (below 90%), and getting worse. GitHub’s leadership blames a ~3.5x increase in service load over the last two years for the degradation – or it might be self-inflicted. We look closer in this week’s bonus issue: blog.pragmaticengineer.com/the-pulse-ai-l…

This is the place to be!

Coming for Pepe’s! 🤣

Author of Designing Data-Intensive Applications, Martin Kleppmann, on why scaling down is just as challenging as scaling up: “Part of what is interesting about modern cloud services, and backend services in general, is how they've introduced the idea of horizontal scalability and shared nothing systems. We can build systems that are able to cope with very high load, even if the individual components are just fairly cheap commodity machines. The scalability story is not just scaling up, but scaling down as well. How do you run a service in such a way that if it has a very small amount of load, it's really cheap to run? That's sort of the same question as ‘how do you continue running a service if it has a very high load?’ Generally, you want the cost and the computing capacity to be roughly proportional to the load that you have and at the low end, that means actually being able to scale down to something that is extremely cheap to run. That's not necessarily a given. It's something that is hard with on-premises software because if you've got a physical machine, that is a unit of deployment. Yes, you could carve it up into two dozen virtual machines, but it still requires some sort of resource allocation. And so, part of what's interesting about some serverless systems, is their ability to scale down and say ‘okay, if you're going to handle just three requests per day, that's just fine as well’”.

@aliceisplaying This is where my council spends the money.

council tax is so funny you move here you rent a place and think well okay i need to pay for electricity and water and maybe gas plus the internet. but no! you'll also pay a bunch of money to the local council. for reasons. i haven't seen this system anywhere else

Uncle Bob making a comeback in 2026 was not on my bingo cards

We’ve identified a security incident that involved unauthorized access to certain internal Vercel systems, impacting a limited subset of customers. Please see our security bulletin: vercel.com/kb/bulletin/ve…

This is the 4th design that I’ve seen on X that looks eerily similar from Claude design in the last couple hours.

Introducing Claude Design by Anthropic Labs: make prototypes, slides, and one-pagers by talking to Claude. Powered by Claude Opus 4.7, our most capable vision model. Available in research preview on the Pro, Max, Team, and Enterprise plans, rolling out throughout the day.

How will AI change software engineering and what skills we value for software engineers? Thuan Pham (Uber’s first CTO, now CTO at Faire): #1 - It’s not the first time software engineer skills have been abstracted “These changes are the most fascinating I've ever seen, including the internet. I remember when we first learned how to do programming we had to know a lot about machine architecture. We have to know about virtual memory and then we have to learn how to write syntax and coding. All of that stuff has been abstracted away now. So with AI, you say, I want X, Y, and Z, and it should be this way and the whole thing gets constructed” #2 - AI is another abstraction that elevates the playing field for devs “It elevated the level of the playing field where people who don't even know how to program can now create decent code that looks good on the surface. So it is game changing, right? It elevated the playing field. Now then in that level of abstraction, how do you tell the great engineer from the good engineer?” #3 - Great engineers continue to be “great,” now as well “From what we see so far, the great engineers are still finding ways to leverage this and accelerate the output even more. They're more inquisitive, they're at the bleeding edge more, they're more innovative. And then there are people who like, okay, well here's the tool that you give me. I'm going to be 2x more productive because I'm using this tool. It's great. But the great engineer continues to break new boundaries. Curiosity. Fearlessness. Willing to innovate, willing to stretch, willing to try new things and break new ground. All of those traits still exist.”

Seen a lot more heckling from the patrons this week than ever! Caddies having to get involved multiple times is not a good look for Augusta. #masters

Give him the green jacket. Call it a day, I can have my weekend back.

Typical sky sports golf commentator bullshit. When liv does bad they pounce on the lower standard of golf they play. Forgetting Hatton who plays in the same league scored lowest round of the day…

A successful night at the aquarium.

Mythos please make this 100% uptime all the time. Thx

Looking forward to attending the coolest/craziest AI event at the Aquarium. It’s gonna be a good one!

I have lost my AirPods. Not the case the individual AirPods FML. 🤦‍♂️

So why isn’t there a Ladies Masters event? I’m sure it’s asked every year when ANWA’s final round is on.