Forget AGI, The Future Is Small

I came across a post earlier today by John Davies, CTO at Incept 5, arguing that the future of AI may not be in the cloud but on your desk.

His point was that while the industry pursues ever-larger models hosted in billion-dollar data centres, Apple is proving that intelligence can be local. The new M5 chip runs open-source models faster, more efficiently, and with complete privacy.

It made me think about where the real opportunity lies. Everyone is talking about artificial general intelligence, about building systems that can do everything for everyone. But I'd argue that the future may lie elsewhere: in tools that deeply understand how we work.

At aethren, we're betting on something smaller, smarter, and closer to the user. We are developing specialised AI models trained on proprietary, domain-specific project data. These models are built to manage projects, not just talk about them: predicting blockers, syncing stakeholders, and getting better with every project they run.

Large, general-purpose models are powerful but heavy. They need enormous resources to run and are detached from the real context in which work happens. In project management, context can make or break you. The challenge is understanding dependencies, timing, ownership, and nuance: all the moving parts that make execution work.

That is why our focus is on small language models trained on domain-specific data. They are faster, more private, and more accurate because they are built for a single purpose: to make real work flow better.

Open-source AI is a powerful foundation. What gives it value is not the size of the model but the quality and relevance of the data. Proprietary project data, refined and structured through use, becomes the true competitive advantage.

The next generation of AI tools will live where the work happens: embedded in the tools teams already use, learning from projects in real time and improving continuously.

While everyone and his dog is focussed on the race toward general intelligence, the quiet revolution is unfolding elsewhere, through smaller, cost-and resource-effective models that deliver practical, immediate impact.