Elyria

We build mathematical models, data science and AI solutions, with complementary software tools that help companies solve problems.

[Elyria @ElyriaX]

We’re live: introducing Elyria!

We started doing various math, coding and business consulting projects back in our university days. Over time, we built the skills and connections to scale what began as a 'side hobby.' But it wasn’t until this past year that we saw the real demand for our work.

That realization came alongside an exponential leap in AI capabilities—transforming not just the speed of our work, but what we could create in the first place. The same effort now goes two, three… many times further, unlocking possibilities we couldn’t have imagined before.

Couple that with our our interest in the coming technological shift, and the opportunity became impossible to ignore. What started as a hobby is now something much bigger. We decided to go all in and build a company around it.

The company is called Elyria, and we’re just getting started.

Check out our website (in bio) for more info.

[Kasper Lahtonen @KasperLahtonen]

To continue with the tradition of posting our numbers: @ElyriaX did 39,392.87€ of revenue in April, roughly double what we did in March.

It was a good month, with some of us stretched quite thin—but not everyone. So there’s plenty more room to run, even with the current setup.

With two larger projects wrapping up by mid-May, and the European summer right around the corner, I doubt we’ll top this number again soon. May might come close if everything falls into place, but after that we expect two quieter months. June will coast on afterburners, and July will be vacation time whether we like it or not (we like it). Then, perhaps, a fresh acceleration starting in August?

It’s good to have grand plans and think big, trust me, I do. But sometimes it’s acceptable, even advisable, to simply keep your head down and double down on what’s already working. I think this is one of those times. And honestly, I quite like it.

In the short term, nothing has changed. As for the long term… I have a plan. Trust me. Again.

[Kasper Lahtonen @KasperLahtonen]

We officially launched @ElyriaX in January. Here’s our Q1 revenues followed by a short update:

January: 11,280€
February: 7,520€
March: 19,417.50€

We moved into our new office in March, and since January and February were mostly leftovers from projects before the launch, I’d say March is when things really started for us. So, we’re only a little over a month into the real beginning, but with April tracking for growth and new projects in the pipe, it seems safe to say that whatever we’re doing is working.

That said, we’re a consulting business and charge by the hour, so an exponential revenue spike isn’t happening anytime soon. Doesn’t mean we’re not looking for leverage—we always are.

In the short term, that means limit-testing ourselves and exploiting the AI loophole (sounds bad, but really just means it’s still easy to exceed expectations before people catch up with the new reality).

In the long term… who knows? I kind of subscribe to the idea that things have a way of working out. There are times when it’s not that simple, but why ruin a perfectly good philosophy with caveats.

I’ll try to share updates more regularly, maybe monthly now that we’ve found some rhythm. Let’s see what echoes back when I shout our numbers into the void.

[Elyria @ElyriaX]

What do you do when your company has a goldmine of data but no clear way to use it?

MV-Jäähdytys, Finland’s largest maintenance company specializing in installing, repairing and servicing refrigeration equipment, had years’ worth of valuable data—purchases, customer orders, employee shifts, suppliers, inventory. The challenge? Turning all that information into real business value.

That’s where Elyria came in.

We helped them make sense of their data to improve procurement, pricing, customer profitability, and even employee incentives. The result? Smarter decisions, better efficiency, and a stronger business.

Want to see what we uncovered? Check out the full case study here: https://t.co/pXxid54HYe

[Elyria @ElyriaX]

For most companies adopting AI, it's all about internal tools to increase leverage, selling more existing products/services with fewer inputs. It's that simple. No need to overthink it.

Elyria is here to help companies create those tools.

[Elyria @ElyriaX]

Things will get interesting. The coming leaps are something you cannot ignore.

[Andrej Karpathy @karpathy]

Projects like OpenAI’s Operator are to the digital world as Humanoid robots are to the physical world. One general setting (monitor keyboard and mouse, or human body) that can in principle gradually perform arbitrarily general tasks, via an I/O interface originally designed for humans. In both cases, it leads to a gradually mixed autonomy world, where humans become high-level supervisors of low-level automation. A bit like a driver monitoring the Autopilot. This will happen faster in digital world than in physical world because flipping bits is somewhere around 1000X less expensive than moving atoms. Though the market size and opportunity feels a lot bigger in physical world.

We actually worked on this idea in very early OpenAI (see Universe and World of Bits projects), but it was incorrectly sequenced - LLMs had to happen first. Even now I am not 100% sure if it is ready. Multimodal (images, video, audio) just barely got integrated with LLMs last 1-2 years, often bolted on as adapters. Worse, we haven’t really been to the territory of very very long task horizons. E.g. videos are a huge amount of information and I’m not sure that we can expect to just stuff it all into context windows (current paradigm) and then expect it to also work. I could imagine a breakthrough or two needed here, as an example.

People on my TL are saying 2025 is the year of agents. Personally I think 2025-2035 is the decade of agents. I feel a huge amount of work across the board to make it actually work. But it *should* work. Today, Operator can find you lunch on DoorDash or check a hotel etc, sometimes and maybe. Tomorrow, you’ll spin up organizations of Operators for long-running tasks of your choice (eg running a whole company). You could be a kind of CEO monitoring 10 of them at once, maybe dropping in to the trenches sometimes to unblock something. And things will get pretty interesting.

[Elyria @ElyriaX]

”Just assume that by end of 2025, at minimum, we'll have multimodal agentic phd AIs doing complex tasks for you on your computer at o1-mini speeds. I highly recommend people update their worldviews and company plans to assume this. This is not hype.”

Act accordingly.

[Will Bryk @WilliamBryk]

I spoke to / heard from a few senior AI labs people this weekend and I now feel even more confident in these predictions from a month ago.

Hearing these same ideas come out of their mouths was quite strange in how real it became. There is a sequence of future events that if you think about from first principles seems inevitable.

We really do now have a test-time compute paradigm that is a straight shot to AGI (no matter how you define AGI). The bottleneck is gathering good data for RL and then all the little product details like improving speed, integrations into our workflows, etc.

Just assume that by end of 2025, at minimum, we'll have multimodal agentic phd AIs doing complex tasks for you on your computer at o1-mini speeds. I highly recommend people update their worldviews and company plans to assume this. This is not hype.

The world will still feel similar bc it will take long for Bank of America and the government of France and everyone to integrate AGI and there will be a compute shortage. But the AGI will exist nonetheless.

Robotics will also take a bit longer and timelines are a bit less clear there but I also talked to people who are doing robotics, and seems we're only talking a couple years later for fully functional humanoid robots.

I believe there is a major gap in our information ecosystem about these topics. There are a couple thousand researchers and engineers thrusting us into the next era of humanity at breakneck speed, but they're not allowed to talk about it bc of NDAs.

This is the most important story of the century -- far more important than covid -- yet there is far less media and discussion about it, beside hype tweets from AI influencers and coy tweets from AI researchers.

That's because there are very few people who:
a) understand how these systems work
b) are able to suppress the "this sounds too crazy" part of their brain
c) are able to suppress the "go full hype" part of their brain
d) are in SF so hear things on the ground beyond twitter/the news
e) actually care about helping the world and are not just trying to sell you something
f) recognize the gravity of the situation and treat it as such
g) are reasonably competent at converting thoughts into words

I believe I am one of these people. As a person who's been running an AI company that trains real models, but one that is not an AI lab (so I can say whatever I want), and someone who enjoys writing about things that are helpful, I believe I can bring an important perspective to this quiet but immensely important conversation.

I feel a duty to write more on X, and plan to do so. You have my word that I will always strive to be as correct as possible and will not write things with hype, though I won't be scared of being provocative.

I also highly encourage anyone else out there who passes the criteria above to write more, create more. The world is desperately in need of media about the world that is coming.

We need realistic AI short stories, new films beyond Terminator, mocked video demos of what future systems will look like.

We all need visceral examples to shake us out of apathy so that we don't just predict the AGI but feel it. That way we can prepare accordingly.