Tomás Mayr

@zeeg @cvxluo our org is interested! Can you shoot me a dm so I can send you the org name?

We're running a new experiment at Sentry that will identify the root cause of issues and create a meta-group of these issues together. This is a true behavioral root cause rather than a traditional source-root. If you have a noisy issue stream that this kind of thing sounds great, you're willing to give helpful feedback, @cvxluo can opt you into the beta!

@boristane Ok but how do I move up the list? ;p

tried this it sucks try nominal.dev instead

People of pi. BIG NEWS. I've sold out. Let me know how you feel about this in the comments below. mariozechner.at/posts/2026-04-…

we do the same thing except we make fun of whoever is #1

@conductor_build any way of using Pi as the harness?

It feels like everyone at work is DDoSing each other with AI. Everyone's capable of slightly more output now. Not better or worse, just more. Which means more input for everyone else. AI is good for producing. We don't have great workflows yet for consuming.

Many people assuming I meant job loss anxiety but that's just one presentation. I'm seeing near-manic episodes triggered by watching software shift from scarce to abundant. Compulsive behaviors around agent usage. Dissociative awe at the temporal compression of change. It's not fear necessarily just the cognitive overload from living in an inflection point.

Unpopular opinion: Unless you’re just prototyping, you should aim to understand as close to 100% of production code generated by LLMs. Yes, all of it. Effective mental models are still important for humans to sustainably maintain and evolve a codebase via prompting alone.

@mitsuhiko Cursor mainly because we have licenses ready and with the modes it has it’s good enough to either work on the codebase or ask questions or use another model if Opus is not doing well. Is using Claude Code much better?

If you are using the Sonnet or Opus 4.5 for coding, but you are not using Claude Code, I would be curious what you use instead and why (eg: amp, Cursor, …)

Here's a bit of coding agent magic fairy dust for the day: Don't tell the agent to do something. Tell the agent to _begin_ doing something. The LLM will, on average, make wiser choices about how much work to bite off.

A question that lives rent free in my head...

@mitsuhiko More or less same experience here on my 13 pro, the lag is unbearable with any transition… but to be fair I’ve had similar experiences with betas in previous versions, fingers crossed.

I really hope Apple did not give up on iOS 26. There haven’t been meaningful improvements to the constant lag, crashes, or bad battery life in any of the updates. If anything it’s getting worse. My iPhone 14 Pro is hot and doesn’t last half a day, despite having a new battery.

@mitsuhiko Heard @withgraphite might be good, but no real experience with it.

What’s your preferred PR/code review AI bot and why? Want to try some alternatives to Claude.

Tech lead vibes

@mitsuhiko I just followed you over the other side. But yeah I do follow your content and enjoy it quite a bit!

I'm quite curious what percentage of the people following me here are actually a) still using X/Twitter b) still interested in my content c) would move to another platform. I'm suspecting that there is a very significant number of inactive accounts behind my followers.

@mitsuhiko Thank you for sharing that! Will definitely give it a read.

@tmayr Austria makes it very hard to integrate due to legal restrictions but it has high amounts of immigration. I wrote about the challenges of the country a while back: lucumr.pocoo.org/2024/2/26/aust…

Sounds about right. The failures in the country that lead to that however are more interesting. In general though it really just comes down to people outside of cities being afraid of immigration.

People will be like, “generative AI has no practical use case,” but I did just use it to replace every app icon on my home screen with images of Kermit, soooo

@ImSh4yy The matrix is bad on the output of detect changes? How did you end up doing that 👀?

Much better now.

@VicVijayakumar We use pgboss for the queue, the worker has been in the same fargate container in the past and now in its own service and auto scales based on jobs in the queue. We also had event bridge triggers to Ecs to spawn another container to do a specific job on demand (before pgboss)

I'm curious to know how you handle asynchronous / background processing in your applications. I have a service running on AWS Fargate that emits messages into SQS. I need to do things based on the payload of those messages. Example: { type: 'compress-image', payload: { bucket: 'big-images-dev', key: 'asdf123.png' } } The SQS is configured to be an event trigger for a Lambda function. The Lambda function then POSTs that message to a webhook in my app with an auth token. The webhook is just a bunch of switch cases which invoke different internal services depending on the message type. When everything used to run on stateful infrastructure (ahem $5 VPSes), I ran scripts that long-polled SQS instead of this.