From Instruments to Intelligence: Scaling ASM Across the Lab
Session overview
Despite significant investments in artificial intelligence across the pharmaceutical industry, laboratory data operations remain largely manual, inconsistent, and fragmented. Data frequently lacks standardized formats, sufficient context, and interoperability across systems. Even when data meets FAIR principles, it often falls short of enabling portfolio‑level insight, site‑to‑site comparisons, or effective root‑cause analysis. These challenges limit how data can be used and slow both scientific and operational decision‑making.
In this presentation, Chris Siegler (MSD), Vinny Antonucci (MSD), and Wolfgang Colsman (ZONTAL), describe their approach to scaling the Allotrope Simple Model (ASM) to standardize laboratory data and semantically connect concepts, relationships, and entities—making information accessible and usable for AI‑driven analysis.
A critical element of this journey is the use of an ASM converter factory to enable rapid scaling across a highly heterogeneous laboratory environment. Traditional approaches, building a bespoke converter for each instrument, can take up to 28 weeks per implementation. By adopting an AI-powered factory of converters, MSD and ZONTAL were able to reduce the time required for the development and delivery of each new converter from 26-28 weeks to approximately 13 weeks, achieving approximately 2× faster delivery.
By treating instrument families as repeatable patterns and industrializing converter generation, validation, and onboarding, MSD and ZONTAL transformed integration from bespoke projects into scalable infrastructure. This approach effectively removed the integration ceiling, eliminating bottlenecks on how many diverse instruments and devices can be onboarded.
As a result, labs can onboard entire instrument fleets rapidly, and move from static data products and delayed insights to real‑time intelligence, enabling scientists to ask new questions, trace results end‑to‑end, and receive answers in seconds rather than days.
