I used to think catastrophic memory loss was something to fear as I got older-an inevitable sign of age. But in the world of AI, it turns out to be something far easier to fix. One of the real privileges-and joys-of being a founder and CEO of a small start up is the freedom to respond swiftly to both challenges and opportunities, without being slowed down by endless meetings or misaligned stakeholders. When you're building something new, speed and clarity of vision matter more than consensus. And let's face it: in 2025, when the entire IT industry is organising itself around lightning-fast execution and AI-enabled development, moving quickly isn't a differentiator anymore.
It's a baseline expectation. As part of the platform we're bringing to market, we've built orchestrated AI frameworks supported by a team of expert-trained, small locally hosted LLMs. For a long time, training-or more accurately, fine-tuning-these models felt like a black art. But after some experimentation (and more than a few late nights), I found a reliable, repeatable approach to preparing our agents with the domain knowledge they need to perform an increasingly wide range of complex tasks. We got here not because we had a perfect plan from day one, but because we were willing to move fast, break a few things, and take some short-term pain to build something truly valuable.
Now, thanks to our AI trainer, fine-tuning is practically effortless. What once took days-sometimes weeks-is now completed in hours. I can't begin to quantify the time we'll save in the year ahead. Ultimately, the ability to identify a problem, respond to it quickly, and then move through it-rather than around it-is no longer just a useful skill. In modern software development, it's a prerequisite.