Convoke

Today we’re releasing the first results from our biopharma knowledge work benchmark. Much of the effort to benchmark AI in the life sciences so far has gone towards research and development (primarily drug discovery) use cases, but relatively little towards evaluating model capabilities in activities that support the commercial, development, or operational functions of a biopharmaceutical company (the customers we serve at Convoke). Our benchmark aims to help fill that gap. We evaluated leading LLMs on six tasks, including target safety assessments, clinical benchmarking, catalyst calendars, and commercial forecasts. Under our evaluation setup, frontier OpenAI models performed best overall. Opus 5 led on target safety assessment reports, but OpenAI models were stronger across the benchmark as a whole. We could not complete evaluations of Opus 5.5 or Fable because of biological-content restrictions. These are initial results. We plan to expand the benchmark to cover hundreds of biopharma work products in the near future. Read the methodology and detailed results on our blog: https://lnkd.in/gyHeHAnV

We’ve extended the functionality of our free program tracker with the ability to create agents that monitor drugs, targets, and organizations of interest and report back on findings automatically. Try it out at platform.convoke.bio, new users get some starter credits. We built this because we often hear from biotechs that they don’t want to have to manually check and reconcile across sources for competitor updates on a regular basis. They want a system that is proactive; that alerts them when something material to their business has changed. However, existing newsfeeds and alerting tools are noisy and spammy. We solved this with LLM layers to gather, deduplicate, verify, and prioritize information across a variety of sources. Biotechs and pharmas are using it to track patents against targets, keep on top of the publication landscape, monitor competitor clinical trials, and more. This infrastructure lets us create highly tailored and customizable monitoring agents for our enterprise customers. We are also excited about finding ways to use agents to bring data to where our customers work, so we’re coupling this launch with Slack and Teams integration which pipe findings into designated channels. Reach out to contact@convoke.com if you’re interested in exploring how your organization could benefit from automated monitors.

Today we’re launching a free, structured data browser for the global drug development pipeline as both a web application and MCP. We first started working on this to improve the performance of our agents on drug landscaping tasks, and after we found it to be useful, we did a 2 week sprint to turn it into a tool that the wider community can use as well. Read more about the drug program tracker on our blog post (https://lnkd.in/gRz_6dcE), or check it out directly by creating an account at platform.convoke.bio. --- The data generation is entirely automated, and no automated system (and no conventional database either) is perfectly accurate. For instance, we know there are gaps in preclinical coverage. We will continue improving the data and the underlying extraction systems. We also hope the community will help us turn it into a better shared resource by leaving feedback in the platform. Corrections will help us improve both the dataset and the systems that maintain it. If you’re in SF, we are sponsoring a hackathon on August 13th where we can help you build agents with our MCP. Register at our event page here: https://lnkd.in/gSwzDj9y If the data could be useful in your product—or you want to build agents that help your team accelerate drug development—get in touch at contact@convoke.bio.

"The industry today, it's like the space industry pre-SpaceX, where every launch is its own project and it's a bunch of throwaway work and you don't learn from that work in a compounding way." In the 4th and final part of The Biotech Startups Podcast conversation with Convoke Founder Alex Telford, that analogy frames the full founding vision: building the operating infrastructure that lets drug development organizations compound their knowledge rather than reinvent the wheel on every program. Here's what they cover in this final installment: - Convoke's platform architecture in practice: building validated, structured datasets from trial registries, regulatory filings, and patents; stitching them together with workflow tools for competitive intelligence and document generation; and embedding closely with customers to encode their specific processes. - How Convoke's engagement model differs between large pharma and smaller biotechs, where big companies have the data but face a veto-cracy that constrains scope, and smaller biotechs often grant one person the latitude to just go through the organization and do things. - A concrete example of the work Convoke is doing: replacing the manual consultant-driven process of refreshing 200 oncology indication profiles every few months, an exercise covering patient populations, competitive landscape, standard of care benchmarks, and emerging clinical data. - The roadmap from reactive data architecture to proactive predictive intelligence, where the system doesn't just store what an organization knows but uses that knowledge to recommend what it should do next, from site selection to trial timeline estimation. - Alex's advice to his 21-year-old self: the world is more malleable than it looks, and if the opportunities in front of you don't seem good, you can create new ones. It's a grounded, specific look at what it takes to build a company at the frontier of AI and biopharma. Head over to Apple Podcasts or Spotify for this episode and more stories only a biotech exec can tell! #biotechnology #lifesciences #podcast

"Signal-to-noise ratio, I think, ends up being much more important for getting good performance out of LLM-based systems than just the raw amount of data." In Part 3 of The Biotech Startups Podcast conversation with Convoke Founder Alex Telford, that principle runs through everything, from why most pharma data is harder to use than it looks, to how tacit knowledge disappears when experts retire, to the writing and prototyping period that eventually nucleated the idea behind Convoke. In this part, they discuss: - Why rearchitecting a pharma organization around AI is less like flipping a switch and more like the factory retooling that followed electrification, where efficiency gains don't materialize until you redesign the whole system, and why some incumbents simply won't make that transition. - The challenge of encoding tacit knowledge before it walks out the door at retirement, and why getting employees to write a flowchart of their own job turns out to be much harder than it sounds. - The two paths Alex sees for genuinely disrupting incumbent pharma: breakthrough target discovery science with a differentiated data flywheel, or a direct-to-consumer commercialization model, and why competing on operational software efficiency alone is not the lever. - Why signal-to-noise optimization produces far better results from LLM-based systems than ingesting maximum data, and the Renaissance Technologies analogy for why simple models on very clean, highly selected data beat complex models on noisy inputs. - His cold-turkey exit from consulting, the period of writing and building prototypes in public, and the realization that the underlying problem was not a collection of point applications but a fundamental rethinking of how data is architected at organization scale. This episode is one of the most substantive breakdowns of AI's real role in biopharma you'll find anywhere. Head over to Apple Podcasts or Spotify for this episode and more stories only a biotech exec can tell! #biotechnology #lifesciences #podcast

"People are generally looking for opinions, not necessarily just information." In Part 2 of The Biotech Startups Podcast conversation with Convoke Founder Alex Telford, that insight from his early consulting years sits at the center of a conversation about seven years of biopharma strategy work, the coordination tax burdening large pharma organizations, and the GPT-3 moment that made everything shift. Tune in for: 🛰️ Landing at a boutique consulting firm in Lucerne only to find himself inside a corporate acquisition two weeks in, and how that Swiss satellite office's unusual autonomy inside a larger company gave him seven years of substantive work across development strategy, regulatory planning, and competitive intelligence. 🫁 His first major project in lung cancer, navigating PD-1 cutoff decisions and combination strategies, and learning that clients don't want an information dump; they want a clear recommendation they can push back on. 🏢 A frank breakdown of why large pharma organizations are so hard to move, from siloed functions each optimizing for their own slice to the unknowability of drug development, where you can always find a preclinical monkey study to justify killing a program. 🤖 How AI and LLMs could help flatten pharma's coordination tax by replacing one-to-one information passing with many-to-one aggregation through a central knowledge system, and what gets lost when drug experts retire with thirty years of tacit knowledge still in their heads. 👨🏻💻 The GPT-3 API moment that triggered what Alex calls his AI psychosis, years of NLP interest colliding with the first language model powerful enough to make him question whether the manual, PowerPoint-heavy work of consulting was the future. This conversation covers ground from Swiss consulting rooms to the foundational frustrations behind what Convoke is being built to solve. Head over to Apple Podcasts or Spotify for this episode and more stories only a biotech exec can tell! #biotechnology #lifesciences #podcast

"I like big ideas. And synthetic biology has a lot of big ideas, like artificial cells and artificial robots. And can you reengineer the chemical supply chain so they use microbial fermentation rather than chemical synthesis? It's all very exciting stuff." In Part 1 of my conversation with Alex Telford, Founder of Convoke, we trace the origin story of someone who followed the biggest ideas in biology all the way from science magazines in Switzerland to a synthetic biology master's at UCL, and what he found when the big ideas met the reality of the lab. Here's some highlights from our conversation: 🌍 Growing up across Europe as an expat kid in international schools, losing best friends to relocations every few years, and how that built a comfort with impermanence that made leaving a stable consulting career later feel a lot less scary. 📚 A childhood subscription to New Scientist and National Geographic, a few well-timed high school teachers, and the very practical decision to study biochemistry at University of Bristol because he couldn't choose between chemistry and biology, got into Oxford, and went to Bristol anyway. 🧫 His undergraduate lab experience spending months on Western blots, finding satisfaction in mastering the technique but confirming that the slow, hard-to-debug pace of bench science was not where he wanted to spend his career. 🔬 Joining a synthetic biology master's at UCL after an email blast caught his eye, drawn in by the promise of engineering living systems, only to discover a field full of cool ideas but plagued by genetic circuits too fragile and unstable to be genuinely useful. 🏭 The realization that pharma was the only real home for someone who loved applied biology but didn't love research, and discovering that consulting even existed as a career path only after he'd already graduated. This one sets up everything that comes next in Alex's story, and it's a great listen for anyone who's ever felt the gap between what a field promises and what it actually delivers. Apple Podcasts & Spotify links in the comments below! 👇 #biotechnology #lifesciences #podcast