July 17, 2026
Big claims, tiny receipts
The Isomorphic Labs Drug Design Engine unlocks a new frontier beyond AlphaFold
Google’s drug AI says it’s miles ahead — commenters say “show us the receipts”
TLDR: Isomorphic Labs says its new AI can predict how future medicines might work better than AlphaFold 3, a major claim in the race to speed up drug discovery. Commenters weren’t swooning though — they wanted hard proof, public access, and more than glossy promises.
Isomorphic Labs just dropped a very big claim: its new AI system, IsoDDE, can help design medicines on a computer better than AlphaFold 3, the famous model that shook biology. The company says it can predict how drugs fit onto proteins far more accurately, spot hidden “pockets” where medicines could bind, and even beat older gold-standard methods while being faster and cheaper. In plain English: they’re saying this could make finding new treatments much easier.
But in the comments, the real action was less “wow, medical breakthrough” and more “okay… and where’s the proof?” One of the loudest reactions was frustration that the post was packed with huge promises but short on details. People kept asking the same spicy question: what exactly changed? Was this a true breakthrough, or just a bigger secret model with nicer marketing? Another commenter twisted the knife with a sarcastic “Thank gods for capitalism!” after pointing out that drug discovery tools tend to stay locked behind closed doors.
And then came the classic internet pile-on: some users said the announcement felt stale, with blunt replies like “This is old news.” Others noted there’s no public demo and no peer-reviewed paper yet, which turned the vibe from celebration to side-eye. Even the nerdier commenters brought receipts from past competitions, basically arguing that flashy AI announcements don’t always win in the messy real world. The mood? Equal parts hype, suspicion, and popcorn-worthy skepticism.
Key Points
- •Isomorphic Labs introduced IsoDDE as a unified computational drug-design system intended to extend beyond AlphaFold 3.
- •The article claims IsoDDE more than doubles AlphaFold 3's accuracy on a difficult protein-ligand structure prediction generalisation benchmark.
- •Isomorphic Labs says IsoDDE predicts small-molecule binding affinities more accurately than gold-standard physics-based methods at lower time and cost.
- •The system is also described as identifying novel binding pockets on target proteins using only amino acid sequence input.
- •The article frames IsoDDE as addressing a remaining challenge after AlphaFold 3: generalising predictive models to novel biological and chemical systems for real-world in silico drug discovery.