Biomedicines VP on technology investments for quality data in drug development | BiotechTV posted on the topic | LinkedIn

LinkedIn respects your privacy

LinkedIn and 3rd parties use essential and non-essential cookies to provide, secure, analyze and improve our Services, and to show you relevant ads (including professional and job ads) on and off LinkedIn. Learn more in our Cookie Policy.

Select Accept to consent or Reject to decline non-essential cookies for this use. You can update your choices at any time in your settings.

View organization page for BiotechTV

30,199 followers

1w

𝐁𝐞𝐧𝐜𝐡𝐭𝐚𝐥𝐤 𝟐𝟎𝟐𝟓: Generate:Biomedicines VP of Informatics Stephen Kottmann describes the investments in technology that ensures only the highest quality data returns back into its models He walks us through the design, build, test, learn model that is the foundation of Generate's approach to drug development. Full video: https://lnkd.in/gMMRVEcY BiotechTV's coverage of Benchtalk 2025 is brought to you by Benchling. BiotechTV is brought to you by: Mispro | Breakthrough Properties | HSBC Innovation Banking | Allucent | NYSE | Klein Hersh | Savills North America | Genezen

2 Comments

Transcript

Sure. So from the get go generates goal has been to use generative AI to really change the way that we do drug discovery, changing it from something that is very sort of a chance oriented process to one that's engineered, predictable, reliable, repeatable and very fast. And so our company really lives on this design, build, test, learn cycle. So everything starts with the design of what we call the variant set. So this is kind of where you get into some of those numbers, a variant set depending on what we are. Are doing and what workflow we're going to go into could be a few 100 molecules, it could be several million. So depending on the workflow that we're going to go into, we design this very large set of sequences and then we flow around this cycle. We take those into the lab, they need to be handed off very seamlessly. And this is where the digitization shows up, right. We don't want to copy paste kind of activities. So we've really invested in building a platform that makes this process flow very naturally from dry lab to wet lab. For those sequences to be made, to be tested and then for that data to flow back into our models pretty seamlessly. Yeah, let's take that into pieces a little bit. So the first piece is that you have to design the labs so that you can test that many iterations of something and that takes very specialized automated equipment. That's right. Yeah. So like I mentioned, we've got a wide variety of workflows that these sequences can go into. Some will go into very highly automated clonal workflows where we are leveraging. Based automation equipment to test sequences and very high throughput. Other workflows use library based approaches where we can test millions of sequences in a single test tube. And then once you've done the testing, obviously collecting the information accurately and analyzing it accurately is super important. Tell us how you do that. Yeah, that's right. So that's a place where we've invested heavily. So we have automated a lot of the data processing we have automated. Agents in our lab to pull the data directly from the machines, so none of our scientists have to walk around with USB sticks and that sort of thing. And then we've built out an asking data collection in QC platform that we use. It supports over 150 different assays at this point. It's very flexible and extensible, but that allows us to capture and and really streamline and standardize the way we collect process QC data so that only the highest quality data is going back into our models. Is there any way to explain maybe in like layman's terms, like what you're doing today compared to how science was done literally just like a handful of years ago, like how day and night different is it today? And what are you able to do today that we could not do five years ago? Yeah. I mean, I think a lot of, I would say falls into two buckets. One bucket is just the approach and the extent to which we have tried to digitize the entirety of our lab process, so. You know, this is where tools like benchling have come in as sort of a key component to our architecture. We've especially leaned on benchling developer tools to be able to build even beyond what they offer out-of-the-box, which has been great. Really the thing that is different about generate though is our upstream models. We are able to start from a place where we our models have already generated good sequences and then we're really just refining and getting to that lead candidate much faster. I have to say this is like a whole deeper discussion, but you know, one thing that our industry is grappling with right now is like cheaper competition from places like China. And I think that automation and efficiency is the answer to that. And the way that the United States not only competes but can stay on top. Yeah, technology and innovation again in our models, but also in the way that we are building out our experimental platform, we invest heavily in these library based approaches and sort of. And cutting edge platform technologies that do allow us to just much more rapidly get through that cycle of design, build, test and learn.

Stephen Kottmann, graphic

Stephen Kottmann

VP, Informatics & Engineering

1w

Thanks for having me on!

Mohammed Waiz, graphic

Fractional CMO @ BioVerse.Social | Helping biotech brands build their socials

6d

Exciting stuff

See more comments

To view or add a comment, sign in

More Relevant Posts

Generate:Biomedicines

47,034 followers
5d
Report this post
How do you build infrastructure that turns data into a strategic asset? Stephen Kottmann, our Head of Informatics, joined Brad Loncar on BiotechTV to walk through the systems behind our design, build, test, learn loop—describing how we generate millions of sequences, move them through automated workflows, and return high-quality data into our models. The goal, as Stephen put it, is to move from “a chance-oriented process to one that’s engineered, predictable, reliable, repeatable, and very fast.” Watch the conversation below. ⬇️ #ProgrammingBiology | #DesignedToScale

BiotechTV

30,199 followers
1w

𝐁𝐞𝐧𝐜𝐡𝐭𝐚𝐥𝐤 𝟐𝟎𝟐𝟓: Generate:Biomedicines VP of Informatics Stephen Kottmann describes the investments in technology that ensures only the highest quality data returns back into its models He walks us through the design, build, test, learn model that is the foundation of Generate's approach to drug development. Full video: https://lnkd.in/gMMRVEcY BiotechTV's coverage of Benchtalk 2025 is brought to you by Benchling. BiotechTV is brought to you by: Mispro | Breakthrough Properties | HSBC Innovation Banking | Allucent | NYSE | Klein Hersh | Savills North America | Genezen
Like Comment
To view or add a comment, sign in
TileDB

14,265 followers
4w
Report this post
As multimodal datasets grow exponentially, the technical complexity of managing terabyte-scale analysis is becoming a significant barrier for research teams. Our survey found that over half of biopharma organizations need platforms that can handle diverse data types without requiring deep computational expertise from every user. 📖 From our latest report: 𝘛𝘩𝘦 𝘚𝘵𝘢𝘵𝘦 𝘰𝘧 𝘔𝘶𝘭𝘵𝘪𝘮𝘰𝘥𝘢𝘭 𝘋𝘢𝘵𝘢 𝘪𝘯 𝘉𝘪𝘰𝘵𝘦𝘤𝘩 & 𝘗𝘩𝘢𝘳𝘮𝘢 2025 Link in comments.
1 Comment
Like Comment
To view or add a comment, sign in
Jeremy Balian

Business Development Leader | Innovating in Life Sciences Drug Target Discovery and Precision Medicine | Driving Revenue through Strategic Partnerships
4w
Report this post
"Life sciences data has always been inherently multimodal. Consider how a single cell can yield data that crosses modalities ranging from transcriptomics to spatial imaging to metabolomics. Zooming out to the level of human health, we can add data types from population genomics, clinical trial results, remote monitoring sensors and EHRs .... .... In this blog post, we review high-level findings from surveys and interviews as well as explore ... [corresponding] issues and opportunities of multimodal data in life sciences. "
TileDB

14,265 followers
4w

As multimodal datasets grow exponentially, the technical complexity of managing terabyte-scale analysis is becoming a significant barrier for research teams. Our survey found that over half of biopharma organizations need platforms that can handle diverse data types without requiring deep computational expertise from every user. 📖 From our latest report: 𝘛𝘩𝘦 𝘚𝘵𝘢𝘵𝘦 𝘰𝘧 𝘔𝘶𝘭𝘵𝘪𝘮𝘰𝘥𝘢𝘭 𝘋𝘢𝘵𝘢 𝘪𝘯 𝘉𝘪𝘰𝘵𝘦𝘤𝘩 & 𝘗𝘩𝘢𝘳𝘮𝘢 2025 Link in comments.
Like Comment
To view or add a comment, sign in
Dan Ryder

Founder & CEO @ Bridge Informatics | Omics Data Management & Analysis
4w
Report this post
Everyone asks: “Which LLM is the smartest?” But if you’re building copilots or domain-specific advisors for drug and diagnostic development, that’s the wrong question. You need models that are: ✅ Secure ✅ Self-hostable ✅ Fine-tunable ✅ Transparent ✅ Built for traceability Here’s the breakdown — including which models are actually being used in biotech R&D today: 🎧 Listen to the 90-second explainer: https://bit.ly/42DZM7P #InformaticsInsight #Bioinformatics #LLM #DrugDiscovery #BridgeInformatics #AIinBiotech

04 - Smartest LLM? Wrong Question. Here’s What Biotech CSOs Really Need

https://www.youtube.com/
Like Comment
To view or add a comment, sign in
Cosmetics & Toiletries

62,831 followers
6d
Report this post
New Wave Biotech and the UK's Center for Business Process Innovation (CPI) have formed a partnership to address ongoing challenges in #lipidextraction, vital for the cosmetics, personal care, health care and food sectors. Using a hybrid #AI-powered simulation platform, the collaboration aims to optimize both solvent and non-solvent extraction methods and improve efficiency and #sustainability in downstream processing. Learn more here: https://brnw.ch/21wWFQ0
Like Comment
To view or add a comment, sign in
Amélie Manson

Key Account Manager @ ACS International - Representing CAS
2w
Report this post
🚀 Excited to speak at BiotechX on October 6th! I'll be presenting: "AI-Powered Pharma: How Big Data Shapes the Industry." In pharma, AI is only as powerful as the data behind it. From clinical trials to real-world evidence, the quality, diversity, and accessibility of data determine how effectively AI can drive innovation, improve patient outcomes, and accelerate drug development. 📊 Good AI starts with good data—and that means breaking silos, ensuring interoperability, and building trust in data sources. 💬 What do you think is the biggest challenge in making pharma data truly AI-ready? Let’s connect in Basel at BiotechX! #BiotechX #PharmaData #AIinPharma #BigData #LifeSciences #DigitalHealth #DataDrivenInnovation #DrugDiscovery #DataQuality #Interoperability
Like Comment
To view or add a comment, sign in
BiotechTV

30,199 followers
2w
Report this post
𝐅𝐫𝐨𝐦 𝐍𝐘𝐒𝐄: Baird Managing Director & Senior Biotech Analyst Brian Skorney, CFA discusses numerous battleground companies in biotech. He shares his thoughts on Soleno, Sarepta, Lexeo, Enanta, Crinetics, Vertex, Xenon, and more. Full video: https://lnkd.in/grrbxYY2 BiotechTV is brought to you by: Mispro | Breakthrough Properties | HSBC Innovation Banking | Allucent | NYSE | Klein Hersh | Savills North America | Genezen

1 Comment
Like Comment
To view or add a comment, sign in
Alex Zhavoronkov
3w
Report this post
The Tech market feels overpriced and reminds me of 2000/2001 in dot com. The only difference is that today the bubbles are massive companies with hundreds of thousands of employees who are now multi millionaires. At the same time, the Biotech market is bottoming. What I am afraid of is what happens with biotech after the AI wave goes down. In my opinion, we need to build the operational and organizational resilience to sustain the long biotech winter and have a good mix of high-novelty programs with a mix of high-confidence/high commercial-tractability programs. At the same time, we must race toward the pharmaceutical superintelligence (PSI) to be able to operate in maximally-automated mode regardless of what happens to the markets.
33 Comments
Like Comment
To view or add a comment, sign in
BiotechTV

30,199 followers
1w
Report this post
𝐁𝐞𝐧𝐜𝐡𝐭𝐚𝐥𝐤 𝟐𝟎𝟐𝟓: Benchling’s Head of Product - Applications Derek Halliday describes how technology is changing the ‘Development’ side of R&D, and how Benchling has helped Merck modernize its bioanalytical systems in this regard. BiotechTV’s coverage of Benchtalk 2025 is brought to you by Benchling. BiotechTV is brought to you by: Mispro | Breakthrough Properties | HSBC Innovation Banking | Allucent | NYSE | Klein Hersh | Savills North America | Genezen
Like Comment
To view or add a comment, sign in

BiotechTV

30,199 followers

View Profile Follow

Explore content categories