Sequence variants are critical quality attributes requiring monitoring during biotherapeutic process development. Conventional SV analysis remains labor-intensive and often inconsistent across instruments and laboratories due to spectral variability and interpretive ambiguity. We introduce an automated, library-driven Proteome Discoverer workflow using CHIMERYS-predicted peptide spectral libraries. By integrating retention-time and fragmentation predictions, the approach reduces bias, supports objective validation of low-abundance variants, and has the potential to harmonize results across analysts/instruments.

Implementing agentic AI in analytical development requires more than workflow automation—it requires connecting experimental data with the scientific knowledge surrounding it. This presentation describes how Merck is integrating structured analytical data with unstructured information contained in presentations, reports, SharePoint repositories, and collaborative workspaces. Using domain-specific AI agents to reason across these previously disconnected sources, scientists can answer complex questions, accelerate decision-making and enable increasingly autonomous analytical workflows.

As the biopharmaceutical industry seeks to accelerate clinical development, automation and artificial intelligence are transforming analytical workflows from data-generation activities into insight-driven decision engines. This presentation will explore how digital analytics, automated testing platforms, and AI-enabled data interpretation enhance process understanding and accelerate decision-making. By linking attribute knowledge, process knowledge, and control strategies through connected digital workflows, these technologies enable faster development cycles, deeper process understanding, and more efficient advancement of medicines to patients.

AutoFlow transforms flow cytometry analysis by replacing subjective manual analysis with fully automated, statistically validated parameter estimation. Validated on 254 cytometry data files from clinical emergency-department blood samples, it achieved high fluorescence agreement while recovering sparse biologically relevant populations often lost by density thresholds. Its scalable runtime and robust normalization make AutoFlow an impactful analytical intelligence platform for reproducible immune profiling, biomarker discovery, and machine-learning–ready clinical workflows at scale.

This talk examines how next-generation NAM platforms, including microphysiological systems and AI-enabled assays, are advancing human-relevant evidence generation for drug discovery, safety assessment, and translational decision-making. Emphasis will be placed on organ-on-chip innovation, multimodal analytics, fit-for-purpose validation, reproducibility, and regulatory context of use. The discussion will highlight how these technologies can generate interpretable, mechanistic, and scalable data that strengthen confidence in development and regulatory decisions across therapeutic modalities and contexts.

Predictive ML models for biophysical characterization are constrained by data scarcity and systematic measurement variability. AbbVie’s iBeacon, a high-throughput viscosity screening platform, exemplifies this challenge: instrument, batch, and run level effects obscure true molecular signal. A flexible mixed-effects modeling framework partitions these nuisance sources from molecular signals, enabling pooling of multi-source datasets. This approach generalizes across biophysical assay types, yielding cleaner training data and more predictive models for candidate selection.

Findable, Accessible, Interoperable, and Reusable (FAIR) data principles are increasingly important for enabling scalable, machine-actionable scientific workflows in environments where data volume, complexity, and generation speed continue to increase. Benchling Validated Cloud has enabled structured data capture through a unified informatics platform supporting globally unique identifiers and standardized metadata. By leveraging structured FAIR data capture and automated data extraction through Python-based workflows, data application capabilities have greatly increased.

Novartis has a broad and diversified biologics portfolio, where analytics play an important role in technical development and decision-making. This presentation highlights strategic approaches to enhancing efficiency and speed in analytical development through digitalization, standardization, and risk-based decision-making. Examples include streamlined analytical workflows, standardized and digitalized processes leveraging platform approaches, predictive stability concepts that support timely development decisions. By combining simplification with smart risk adoption and digital tools, these elements aim to reduce complexity, accelerate analytical development, and ensure robust, scalable, and reliable support for the expanding biotherapeutic pipeline across diverse modalities.

As automation expands across industry workflows, the bottleneck in early-stage developability screening has shifted from protein production and sample handling to downstream analysis and replicate-heavy plate designs. To address these, we have re-engineered assays to return a single unambiguous output per molecule—exemplified by the HIC slurry assay, with additional assays following the same framework. Every plate is designed with process and range-bracketing controls to capture each assay's sensitivities, enabling a singlicate format with run-to-run comparability across campaigns. Together, these changes support high-confidence developability decisions at the throughput that modern discovery pipelines demand.

Stability is a critical attribute in biotherapeutics research and development, affecting product quality, formulation behavior, and manufacturing outcomes. This presentation highlights bioanalytical and separation methods, along with computational and AI/ML tools, that are used to characterize and interpret stability-related properties in preclinical and early-development settings. By combining experimental measurements with predictive modeling, these approaches provide actionable insights leading to more informed development decisions.

The integrated wet-dry lab platform at Generate: Biomedicines has advanced developability from early risk filtering to multi-parameter co-optimization during lead generation, supporting successful lead nomination across multiple programs and progression into clinical development. This presentation will highlight data-generation and learning strategies that transformed developability from a high-throughput screen into a core design objective, as well as recent steps toward a developable-by-design discovery platform.