August 2024 Summaries

Recursion Pharmaceuticals is working to bring safe and effective new medicine for cancer and other diseases by challenging assumptions and convention, with a focus on efficiency in the drug discovery process. The company's data team has been tasked with bringing speed and accuracy to this process, but was finding it difficult due to inefficient tooling and processes that positioned them as a liaison between scientists and executives. To address this, Recursion reconstructed its entire data process and tooling, adding Hex, which enabled the company to scale up its operations and bring together multi-disciplinary teams to access data from a centralized backend, called the Comprehensive Map of Biology. This new workflow allows for faster testing for success, and has transformed the way the company approaches drug discovery, enabling scientists, data scientists, and leadership to work together more seamlessly and make better decisions. The new system is powered by a hybrid architecture that decouples processing from storage, centralizes all different data types in BigQuery, and uses Hex for complex analysis, visualization, and last-mile delivery of multi-disciplinary data.

Recursion Pharmaceuticals has revamped its data workflow to accelerate drug discovery by transitioning from a linear pipeline approach to a scalable data loop, allowing for more efficient collaboration among its multi-disciplinary teams, including scientists, engineers, and leadership. This overhaul involved the integration of Hex, which facilitated complex analysis and visualization, and the use of their custom-built BioHive supercomputer for data processing. By decoupling data processing from last-mile delivery, Recursion has improved its ability to manage vast amounts of data and make holistic decisions, thereby enhancing the drug discovery process without needing to increase headcount. The new workflow supports a culture of collaboration and efficiency, with centralized data architecture enabling different experts to work together seamlessly, ultimately leading to faster and more cost-effective drug development.

Daniel Sternberg, leading data and AI teams at Notion which recently surpassed 100 million users, shared his philosophies on data team strategy during the first Friends of Data event. He discussed perspectives on data team structure and ROI, balancing data-based decisions with craft-based ones, and one of his most influential books. Sternberg emphasized that while data is important in decision making, it should not be solely relied upon, especially in a SaaS subscription business like Notion. He also touched on the use of AI in their work, noting its potential to accelerate tasks for those with basic data skills but acknowledging that it cannot replace human expertise and judgment entirely.

Data workflows are becoming increasingly fragmented as professionals juggle multiple tools for tasks such as quick queries, operational reporting, data apps, forecasting, model development, workflow automation, and self-serve analysis. This leads to a Tower of Babel scenario where everyone is using different languages and tools, making communication and collaboration difficult. Hex aims to address this issue by creating multi-modal platforms that support various ways of working with data in one integrated experience. The company's "cells" system allows users to combine code, no-code, and AI modalities into a unified workflow, enabling seamless collaboration and creativity within teams.