28th of Sep - 2nd of Oct. This year. Varberg, Sweden.

Tickets are now on sale. The Early Bird started yesterday. It lasted for 3 hours. We're very pleased that people are so excited :)

Those who dare speak will find our Call for Talks.

Seriously though. We want speakers of all experience levels from all parts of the world. Take a look. The previous conference was a blast.

We will not make less

The fundamental learnings about what we need to do to be sustainable as a species, improve the environment, recover sanity in as individuals, as collectives, is relatively simple. Do less. We are overall, in abundance. Do less and do what you do in a slightly better way. This'd turn everything around it seems.

I don't hold a lot of hope for that in the grand scheme. I hold a small hope for that for myself though I struggle to not Just Go.

For software. With LLMs in the mix? We are in for a cambrian explosion of software. Open source will be a morass of unknown-quality repos that claim to do whatever the originator intended it to do and it may or may not do it. This has often been the case in the past as well. But fewer people felt ready to take on absolutely bonkers things. I already see an explosion in UI frameworks, Agent orchestrators and various other LLM abstractions and just a general uptick in libraries. Most but not all intended to serve building more things with the LLMs. This is along the obvious path for a developer with a new tool so anyone with the time and energy will slip down that route.

This will lead to a very confusing world where trust, taste and ecosystem knowledge becomes exponentially more important to know what the heck is there and what is chaff, noise, slop. And there'll even be distinction between useful slop and terrible slop to consider.

I'm in a more narrow space. Embedded Linux type stuff like Nerves. We still see the uptick here and we also see the utility and with some of the client work I have where it makes sense I'm generating new software like no-one is watching. Essentially embedded Linux has always had an enormously high potential ceiling for what could be done but the reality of doing it. Utilizing all the hardware, accelerating all the processing, doing more on-device, tooling up all the things. There have never been enough people who know enough to do it, outside the smartphone manufacturers.

Every chunk of hardware is somewhat bespoke and has specific concerns and each vendor has their cursed tooling and their weirdo accelerators and their sad, sad software ecosystem. You can read this thread on the RK 3588 for some sense of both the inanity of the vendor-provided world and just how much a dev and his bots can progress. Being able to actually spend cycles on getting these things working improves performance, improves energy use, lets us do more with less.

Lots of things are under NDA so the software generated will never see the light of the public. Especially the tooling. But I bet we'll also see people do clean-room re-implementations of stuff with bot help as well to produce open things. Stupid, tedious things that can be verified to work.

This niche is a target-rich environment of hard and tedious things that we can do that is not worth the man-months but may be worth bot-days and man-hours. From my current vantage point I see a lot of upside and not that much downside. 

In the large perspective I expect it to be an absolute mess.

I'm here on email or on Bluesky and the Fediverse if you want to reach me :)

Thank you for reading. I appreciate you.