Lab work is "high mix, low volume," like driving, making it hard to automate. Traditional automation is like a subway: efficient but inflexible. AI enables "autonomous" labs, akin to Waymo cars, that handle the vast variability of experiments, which constitutes 99% of lab work.

The lab of the future should abandon human-centric design and instead emulate a data center. This means prioritizing density, energy efficiency, and automation to maximize the generation of scientific data ('tokens') around the clock, with minimal human intervention.

A key benefit of autonomous labs isn't just speed but perfect documentation. AI-driven systems eliminate human variability—like slight changes in pipetting angle—that is impossible to document but critical for reproducibility. This creates the pristine, detailed data needed for advanced AI models to learn effectively.

The biotech industry often believes its processes require unique, specialized robots. In reality, well-proven robotics from industrial and logistics sectors are applicable. The key is thoughtful system design and adaptation (e.g., sterilization, end effectors), not reinventing core technology.

Despite labs being human-centric, humanoid robots are a poor solution. The primary task is moving samples, which specialized tracks do better. Biology, like chip manufacturing, is a microscopic discipline where the goal is to remove human-scale limitations, not replicate them with robots.

While automation is crucial for ensuring consistent, replicable experiments by eliminating human variability, it risks removing the "irregularity" that can lead to unexpected breakthroughs. This creates a new design challenge: engineering for human ingenuity alongside automated systems.

An automated lab just executes pre-defined experiments at high throughput. A "self-driving" lab, like Radical AI's, autonomously designs and runs entire research campaigns, deciding what to do next based on results, much like a human scientist.

Unlike pre-programmed industrial robots, "Physical AI" systems sense their environment, make intelligent choices, and receive live feedback. This paradigm shift, similar to Waymo's self-driving cars versus simple cruise control, allows for autonomous and adaptive scientific experimentation rather than just repetitive tasks.

A key overlooked parameter in building autonomous labs is 'knowledge'—not just scientific data, but the practical expertise of technicians in handling failures. Without embedding this troubleshooting logic into the system, labs cannot achieve true 24/7 autonomy and will fail to scale cost-effectively.