Building AI scientist that does real world science

I’m working on an AI system designed to close the loop in scientific research.

The problem I kept running into during my PhD in chemistry:
Scientific research is still very manual: humans design experiments, write protocols, run them, analyze results, and decide what to do next. This makes iteration slow and expensive.

The idea:
An AI system that can:

• Read scientific literature

• Generate hypotheses

• Write or adapt experimental protocols

• Analyze experimental results

• Decide the next experiment to run

The goal is to reduce the manual overhead in research and significantly speed up discovery, while keeping humans in the loop where needed.

I’m curious to hear from this community:

Would love feedback from people in biotech, AI, or R&D.