Optimal has been acquired by Source.ag! I never imagined I’d go this deep into greenhouses. I started Optimal after seeing DeepMind cut the energy consumption of a Google datacentre by 40% using Reinforcement Learning. I was part of Shimon Whiteson's RL group at Oxford working on similar algorithms and that result changed the course of my life. If RL could optimise datacentres, why not other industries? A few weeks later, I cold-called Frank van Kleef, owner of one of the largest greenhouses on the planet, and pitched him AI controlled greenhouses. He invited me to spend a few days there to see how they operated. What started as a few days ended up being 9 years. Modern greenhouses supply $30B of fresh produce per year and are critical for keeping vegetables affordable and available under extreme weather and water scarcity. But there are not enough growers to run them. This shortage prevents new greenhouses from being built and leaves our food system exposed. This is the grand challenge Optimal set out to solve. We built AI that directly controls greenhouses through existing hardware: heating, cooling, lighting, shading, thermal screens, CO2 injection, irrigation. We developed hybrid physics and machine learning models of the crop and the greenhouse, and embedded them in an AI controller, to plan the growing environment days ahead and re-optimise in real time as the weather changed. In a commercial-scale greenhouse in the Netherlands we delivered 13% more yield with 27% less energy than the benchmark Dutch grower. Since then, our AI has run commercial operations in the Netherlands and Canada. Joining Source is an important next step in our journey. Source's software runs in more than 500 greenhouses across 25 countries, monitoring the crop, forecasting harvests, and automating irrigation. They hold the world's largest dataset on greenhouse crop growth. Optimal adds control of the greenhouse itself. Together, we close the control loop: observe, orient, decide, act. Growers run 10x more greenhouses, at higher yield and lower cost. This removes the barrier to scale and lets greenhouses be built anywhere on earth, supplying affordable produce under ever more extreme weather and energy volatility. Excited to build the next steps with Rien, Ernst and the rest of the Source team.

Rest In Peace @ScottAdamsSays Thank you for your generosity and your thoughtful words of truth. You were an excellent human being.

Designing an org chart today means starting with AI and then figuring out where the humans fit in

Our open roles in engineering are changing by the week as new AI models are released

Caught up with robotics kitchen founder @josefchen - offering insanely good £ per gram-of-protein ratios on food bowls at @KAIKAKU_AI Built their own robots from scratch enabling them to offer higher quality at a lower price than competitors. Check it out!

Matches our experience. Human expert + AI copilot maximises output. answer.ai/posts/2025-01-08-d…