Research assistant AI has split into two ambitious branches. The first automates the scientific method itself: Periodic Labs ($300M raised) is building AI scientists and autonomous laboratories for the physical sciences, Edison Scientific ($70M) compresses months of research into a single day, and Axiomatic AI ($25M) builds intelligence infrastructure for verified science and engineering. In mathematics, Harmonic ($295M) pursues hallucination-free verified reasoning and Axiom Math ($264M) an AI mathematician that generates and proves new knowledge. The second branch automates commercial research: Listen Labs ($100M) runs AI-moderated customer interviews at scale, and Aaru ($50M) simulates entire populations to predict events.
Users are correspondingly varied — R&D organizations and labs, pharmaceutical and materials companies, forecasting teams, market researchers, and academics. What they share is a bottleneck: expert attention. These tools multiply it by reading, hypothesizing, simulating, or interviewing at machine scale.
The engineering pattern that separates serious platforms from chat wrappers is verification. Hallucinated citations are disqualifying in research, so leaders build in formal proof checking (the approach behind Harmonic), closed-loop experimental validation, or transparent sourcing for every claim. AMI Labs — the category's largest raise at just over $1 billion — reflects a deeper bet: Yann LeCun's lab is building foundational world models aimed at reasoning beyond what language models alone can do.
Buyers should test on questions where they already know the answer, inspect how sources are cited and verified, and clarify who owns discoveries the system contributes to. NeuronFeed tracks 28 research-assistant companies with $2.4 billion in combined funding.