How AI Cut Our Lead Compound Identification Time From 18 Months to 4
A first-person account of how CompoundAI's platform compressed lead identification — and what specific decisions made the difference in the CAI-001 program.
Read more
CAI-007 Safety Cohort Fully Enrolled in Phase I Trial
Our tau aggregation inhibitor targeting early-stage Alzheimer's disease has completed enrollment of the initial safety cohort. Data readout expected Q4 2026.
Read more
The Protein Folding Revolution and What It Means for Small Molecule Design
Structure prediction changed what's possible in small molecule drug design. Here's what actually changed in practice, and where the gaps remain.
Read more
CompoundAI Releases ADMET Prediction Module v3.2
The latest version of our ADMET prediction engine adds support for 12 additional toxicity endpoints and improves metabolic stability predictions across CYP3A4 substrates.
Read more
Why Most AI Drug Discovery Startups Fail at the Wet Lab Stage
The gap between computational output and experimental validation is where most AI drug discovery programs quietly collapse. Here's why it happens.
Read more
Generative Chemistry - Designing Molecules That Don't Exist Yet
Generative molecular design is moving out of the demo phase. Here's what it can do, what it can't, and how we're using it in active programs.
Read more
Our First IND Filing - Lessons From Taking AI-Designed Compounds to the FDA
What we learned preparing an IND for an AI-designed compound — from the pre-IND meeting to the CMC package and the clinical hold that taught us the most.
Read more
CompoundAI Expands Research Collaboration Network to 14 Academic Partners
We have added four new academic research partnerships this quarter, bringing our total collaboration network to 14 institutions across the US and EU.
Read more
The Data Quality Problem in Drug Discovery Nobody Talks About
Drug discovery models are only as good as the data they train on. The quality problems in bioactivity databases are systematic, significant, and mostly ignored.
Read more
Retrosynthesis Planning With AI - Making Impossible Molecules Practical
AI-assisted retrosynthesis planning is changing what medicinal chemists attempt to synthesize. Here's where it works and where it still needs human judgment.
Read more
Why We Partner With CROs Instead of Building Our Own Labs
The decision to use contract research organizations rather than build internal wet lab capacity is strategic, not a limitation. Here's our reasoning.
Read more
Target Identification vs Target Validation - Where AI Actually Helps
AI tools are useful for one of these steps and mostly irrelevant to the other. Understanding the difference is critical before investing in either.
Read more
What Three Failed Clinical Candidates Taught Us About Computational Prediction
Three compounds we were confident in failed in clinical development. Here's where the models got it wrong and what we changed as a result.
Read more