Home /By industry /Science & research
§08By industry

Science & research

For researchers who need structure more than they need shapes.

Scientific research saw two ruptures since 2024: AlphaFold, which made protein-structure prediction trivial, and Co-Scientist-style agents, which automate hypothesis formulation. The job stays manual — but the perimeter of "manual" has changed.

01
Molecular structure
Google DeepMind / Isomorphic Labs
AlphaFold 3

The standard for predicting the 3D structure of proteins, DNA, RNA and ligands. The 2024 chemistry Nobel went to the team behind it.

02
Drug discovery
MIT CSAIL × Recursion
Boltz-2

Predicts structure and binding affinity jointly, 1,000× faster than classical physical methods. Open source.

03
Hypotheses & papers
Google DeepMind
AI Co-Scientist

Multi-agent system built on Gemini 2.0 that generates hypotheses, designs experiments and drafts papers. Confirmed in 48 hours a gene-transfer mechanism that took Imperial College researchers ten years to validate.

04
Literature review
OpenAI
Deep Research

Synthesises hundreds of scientific papers into a cited report in under an hour.

Our editorial verdict

For a research lab, the winning combo remains AlphaFold 3 (free via the DeepMind site) plus Claude for the paper write-up plus Deep Research for the bibliography. Boltz-2 if you are in biotech.