Next week I'll be in Malaga for ElixirConf EU. And I'll be speaking on Nerves, media streaming, cameras and a little bit of inference.

I'm very excited to reconnect with the community in person. I've missed you all since Goatmire 2025.

We are running a Nerves workshop and Elixir Unconference as a collab between Nerves and the Erlang Ecosystem Foundation. So show up on the 22nd for that. Hacking on the Nerves Starter Kit UI and all sorts of stuff.

Goatmire Elixir 2026

28th of Sep - 2nd of Oct. Varberg, Sweden. Sweden's largest, smallest, best and worst Elixir conference.

Tickets

A third of the tickets are gone. We have not announced most of the line-up yet. Be warned that they might move fast as we announce more.

Acquire your passage!

Oh. And we confirmed a speaker I'm very excited about.. read below.

Call for Talks is over. Announcing speaker line-up soon.

Sponsors

If you think your company can and should support the conference we are actively talking to sponsors now. You can just respond to this email.

Workshops

Just added a lot of seats for Sam Aaron's music creation workshop with Tau5.

Check out the workshops if you are attending the conference or are on the Swedish west coast. They are free, not all are Elixir and all of them are phenomenally interesting.

A Hundred Eyes

by Quinn Wilton

"Let it crash" is the BEAM's answer to failure: processes crash, supervisors restart them, and the system recovers. That confidence is earned, but some bugs hide behind it. They live in the gaps between processes, where getting things right is subtle and getting things wrong looks exactly like getting things right. You'd have to never blink to see them.

Argus Panoptes was a hundred-eyed giant who never blinked. I named a tool after him and pointed it at libraries you use in production.

Focus & Intent

This is about LLMs. If you've completely had your fill. Fully understand. It is just what's on my mind and that's what I write about. I would not describe my stance as enthusiastic. I also don't believe this is going away. And LLMs have certainly let me implement things I could not have done solo. Take that as my baseline.

Rabbit holes are a danger for productivity and a boon for creativity. And every so often they are key to unlocking things that you simply could not have done without going deeper than made sense.

LLMs makes everything into a rabbit hole. This is why you see a lot of devs building their own LLM orchestrators now. You sense the potential velocity and you want the overview, the control, the perfect setup. Which is always a mirage of course. But of course there is productivity to earn there. There is also immense risk of losing your mind. In more ways than usual.

Customizing and tooling up your dev workflow has always been a balancing act where pragmatism usually beats both perfectionism and complete disregard.

There used to be cheap heuristics for what was out of bounds for me.

"Oh, that's C. I guess I'm not touching that."

"The driver seems bad. Nothing I can really do there."

"This is fast enough. Making it faster would mean using the weirdo GPU."

"Making this CSS transition and animate would just take too much time to work through properly."

"I'd love a Web UI to watch the progress on this but I can't spend time on that."

When I don't need to know the entire C code base and figure things out myself and rather just review a small change. Doable. If the good solution is unlocked by fixing the driver. Certainly worth an hour of investigation to see if the driver can be fixed. Throwing things onto the weirdo GPU somewhat blind is fine if we can wrap it up in tests to sanity check it. When trying to make something nice takes 30 min for five solid attempts instead of an hour for one. Then it is way more feasible to go an extra mile. And when the web UI is essentially included for free in making the thing.

For someone with my type of creativity where I really enjoy and value building things this is straight up dangerous. I'm very glad I've had about 20 years of practice at focusing on what is important. The heuristics of "I don't know how to do that so that's not on the table" is something that loosens over time, with experience. But recent experiences with LLMs have blown the bounds off towards the horizon. I can barely see them. So direction, focus and intent feels more important than ever. You can do more things, do things deeper, build it out more completely. If you wield it right (and it will fight you) I do believe you can do better work.

A friend who is very excited about AI noted that taste and creativity will be more valuable than ever. And that is fantastic copium for me since I like my little opinionated style and well .. look at my talks, the conference I run, how I do things. I'm not saying my taste and creativity are world class or anything. I'm swedish, that's illegal to think or say. But I do care about that stuff and I do work those muscles continuously. And I think there is something to it.

I don't think most companies give two shits about any of this and I think we'll see more and more stuff break really bad as they just ship more and worse with zero understanding or care.

Focus and intent has always been a struggle for me. That also means I'm aware and constantly thinking about it, evaluating it. And I think that has become way more important.

I'm here on email or on Bluesky and the Fediverse if you want to share. You can just reply to this email :)

Thank you for reading. I appreciate you.