Hey everyone! We’re building KemJet, a collaborative platform for AI-powered drug discovery, and we’re getting ready to open up access ahead of launch.
I’m looking to connect with people who are genuinely interested in this space, especially:
People building specific ML models for drug discovery (ADMET, molecular generation, retrosynthesis, docking, protein modelling, etc.), beyond just LLMs.
Researchers and developers with interesting ideas about how we can approach drug discovery differently.
People actively sharing ideas, research and conversations around computational chemistry, drug discovery and AI on X, LinkedIn and elsewhere.
I’m interested in people who want to contribute ideas, challenge existing approaches and help shape what we’re building. Not just people looking to try another tool.
I’ll be giving early access to interested people so you can explore KemJet, experiment with the workflows and see what we’re working on.
If any of this sounds like you, send me a DM. I’d love to connect and hear what you’re working on.
https://t.co/Ixbedi8oR3
Nigerians are already poor enough, struggling to afford food, medication and basic healthcare. Now, this prolonged power outage is another burden they have to negotiate.
How is a Type 1 diabetic supposed to keep insulin refrigerated? What about a glaucoma patient using latanoprost?
I wonder how many people are in more dire situations because of this. How many have suffered complications, or perhaps even lost their lives? Do you know what it means to have chickenpox in this kind of heat, without electricity or air conditioning? A fate I can only describe as hell on earth.
People are already struggling to stay alive. Why must even that be made this difficult?
Nigerians are already poor enough, struggling to afford food, medication and basic healthcare. Now, this prolonged power outage is another burden they have to negotiate.
How is a Type 1 diabetic supposed to keep insulin refrigerated? What about a glaucoma patient using latanoprost?
I wonder how many people are in more dire situations because of this. How many have suffered complications, or perhaps even lost their lives? Do you know what it means to have chickenpox in this kind of heat, without electricity or air conditioning? A fate I can only describe as hell on earth.
People are already struggling to stay alive. Why must even that be made this difficult?
@KarlGademann In that sense I also think, that using specifically very large language models for this is not cost efficient either and I’m sure we perfectly agree there 😂
I’m not a bio-AI guy who believes ML will solve every problem in biology by just throwing in money for compute or that you can sideline real professionals by using LLMs, but I honestly feel there’s a growing hostility towards A.I in the field, that’s now starting to blind judgement.
Retrosynthesis is important, especially in drug discovery. What’s the point of generating promising molecules if we cannot establish whether they have plausible, accessible synthetic routes? Knowing whether an AI-generated molecule can realistically be made is part of evaluating its usefulness.
Of course, making the molecule is another challenge entirely. But that doesn’t make progress in retrosynthesis any less important. These are complementary problems, not competing ones.
I agree with you, and I think that’s a fair point. I suppose where I see the value is less in solving problems that chemists currently cannot solve, and more in the speed and scale at which we can explore them.
With AI-driven molecular generation, we can potentially generate hundreds or thousands of candidates in a fraction of the time. But generation at that scale also means we need equally efficient ways to assess synthesizability, explore multiple routes, and filter out molecules whose proposed routes are impractical. Otherwise, we’re just accelerating the production of ideas without accelerating our ability to validate them.
I think that’s where AI-assisted retrosynthesis becomes particularly valuable: not necessarily replacing chemists or solving an entirely inaccessible synthesis problem, but making the feedback loop between molecular generation, route assessment and validation much faster, so that chemists can ultimately review a more refined set of candidates and focus their expertise on the final assessment.