Thinking at the Edge

In 2025 we published a reflection that grew out of our ongoing work with small, local AI systems. It began with a simple observation from biology: an octopus does not rely on one central brain. Each arm has its own local intelligence that senses, decides, and acts without checking in with the centre.

This became a useful way to think about the future of AI and about the experiments we run at Gervi Labs. We have been moving more intelligence out of the cloud and into the physical world, placing models inside unexpected objects, including prototypes like a cuckoo clock that responds and adapts on its own.

The project explored the benefits of edge intelligence through three lenses:

1. Environmental impact. Running everything in distant data centres has a cost. Local AI reduces traffic, energy use, and server load by doing more work where the data is created.
2. Decision speed. In settings where milliseconds matter, local processing outperforms round trips to the cloud. This is already visible in medical devices, autonomous systems, and traffic technology, and it shapes the way we design our own prototypes.
3. Digital sovereignty. Local models keep data on site, strengthen resilience, and reduce dependence on large external providers. This aligns with our interest in decentralisation and in giving people more control over the systems they use.

The project framed edge and cloud not as opposites but as parts of one hybrid nervous system. The cloud remains the place for training and heavy computation. The edge is where intelligence acts in real time, close to sensors and people.

Thinking at the Edge captured a moment in our work where metaphor, research, and practical experimentation met. It helped us shape new prototypes and guided our early explorations into distributed AI systems.