Now that it's been over a week (and that I've gone through a few more exams/"quizzes" ðŸ˜), I'm looking back at ATmosphereConf 2026 and thinking of all the cool things I watched and did and talked about with everyone there!
Firstly, it was awesome to find out that a conference for the AT protocol was happening at UBC as it was literally right where I'm studying. It was super nice to be able to go right after classes to join some of the talks or "unconf" events that were happening in the AMS Student Nest.
Then the talks themselves! There were so many... 😠I wish I had a cloning machine that could clone me into 3 people so that I could be in every room at all times. There was always a talk, a unconf event, or some presentation happening in every room. It was super awesome that it seemed like there was so much happening but at the same time it was really difficult to go to all of the sessions. I feel like for the next ATmosphereConf it would be much better if it could be split apart a bit so that there's maybe a few minutes break between some of the sessions to get my bearings and move to the next one.
The events were super interesting!! The first one I attended was the at://advent, an atproto adventure1 workshop on Friday. It was run by and . I loved the idea of having Advent of Code-styled challenges that go through and teach you how atproto works! I wish is had a bit more of an explanation right at the start or for day 1 at least as it got confusing sometimes when I was just learning about atproto and everything to do with it while also solving those challenges 😅. I was able to get through most of the challenges by reading through the links and info they gave and just general screwing around which was great!
The next event was the 2026 Atmosphere Report2. This was a presentation where and talked about the current state and future of Bluesky. It was super interesting to see all the websites and apps built on atproto (such as , , , and many others). Then they moved on to the features and roadmap they had laid out last year which included adding a method for having permissioned data in atproto. This sounds awesome and honestly I can't wait for it to become a fully public thing that everyone starts using! The possibilities truly are endless and I feel that it was one of the main things that stopped people from making certain types of apps and services on atproto (such as a file storing/file sharing service like Google Drive). But now with permissioned data the sky's the limit!!
And then they moved on to Attie3... I have complicated feelings about Attie. On the one hand, it's awesome that they're making it easier for anyone to make any feed (but more importantly, app/website) just by asking an AI agent called Attie. But on the other hand, this might cause issues later down the road when there's a thousand different apps that all do or access the same data. That on it's own isn't inherently a bad thing, but it's the fact that now people can just make apps without necessarily needing to see whether an app with the needed functionality already exists and just take it's code straight up and publish it. It's sort of similar to the comic of creating new standards4 but the new apps are now taking users that could've helped some other app to grow bigger which might be discontinued due to low usage.
Also, I feel like there were more important issues that they could've focused on rather than releasing an AI agent that ChatGPT, Claude (they use Claude Sonnet 4.6 by default by the way), Gemini, Llama, Grok, or DeepSeek - and more recently Qwen - could've very easily done the same job. They could've focused on so many other more important features such as editing posts and maybe sending media in DMs but no instead we have Attie. 😑
ALTHOUGH I WILL SAY, it's nice that they are making the barrier to entry lower so that we can have more types of apps, but this may or may not become a problem soon (as is the case with all "AI" agents/models/chat-bots). As always, we will see soon.
On to something much more exciting, more discussions/panels and talks! On Sunday I attended the How to have more non-english speaking users5 discussion/panel run by . It was super interesting to learn about translating and localising apps to attract a wider audience. For example, I didn't know that only 5.56% of Brazil speaks English6!
There was also the Blousques: Case Study on the Challenges in Translating Bluesky's UI7 presentation hosted by which was a really interesting addition to the discussion! It was super cool to learn about different methods of translating an app's UI elements (user-generated content is supposedly much more difficult). Turns out "user-generated content" is actually not such a simple line to define too as some of the tags on Bluesky's platform are technically user-generated but look like UI elements to users. It's also sad to see that companies won't care about localisation unless given an extremely strong reason to do so which shouldn't be the case at all. Software should be free and accessible for everyone, everywhere.
But overall, the entire event was awesome!! Volunteering there was awesome as I got to learn more about atproto and all the cool stuff being made constantly! Also, apparently more than 3 times the number of people came to the event this year than last year (which if we extrapolate, by 2041 we'll have 4304672100 people at the event). The next ATmosphereConf might be happening in Europe somewhere so that's going to be interesting and very fun! 👀
See you around the internet and hope you all had a great time at ATmosphereConf 2026!