
Create your personal agenda –check the favourite icon
Artificial intelligence and data analytics are transforming how we work, make decisions, and solve complex problems across industries. Yet, the true potential of AI depends not simply on what algorithms can do, but on how effectively humans prompt, interpret, trust, challenge, and act on their recommendations. This talk explores the evolving partnership between AI and human judgment—where AI can augment expertise, where human intuition remains essential, and why people sometimes override algorithmic recommendations. Ultimately, realizing the promise of AI requires finding the right balance between data-driven intelligence and human expertise to make decisions that are faster, smarter, and more impactful.
Learning Objectives:
Senior Representative Ecolab
Senior Representative, Sartorius
Senior Representative, Catalent
Microgravity is usually discussed as a research environment. Eascra is treating it as a manufacturing parameter. Janus Base Nanoparticles — a DNA-inspired, non-lipid delivery platform designed to reach tissue types lipid nanoparticles cannot reliably access — show measurably better structural consistency and therapeutic performance when formulated in orbit than in matched Earth-made controls. Turning that observation into a supply chain is a different problem from proving it.
Under NASA's InSPA Phase II contract, Eascra is developing semi-autonomous production hardware with two independent engineering partners, running coordinated synthesis campaigns aboard the ISS, and returning material for ground characterisation. This session covers what industrialising that actually requires: the hardware and infrastructure needed for autonomous production off-planet, the reagent stability and cold-chain constraints that shape an orbital supply chain, and the regulatory work of framing microgravity as a controllable process parameter under ICH Q8, Q9 and Q10 rather than an exception to them.
Key Takeaways:
I’ve delivered Talent Science™ Boardrooms and keynotes at major global conferences including BPI (Europe & Boston), The Economist, Advanced Therapies, Bio-IT, Bio-Europe and DCAT, and would be delighted to explore how this could enhance the Festival’s strategic and leadership-focused sessions.
Senior Representative, FUJIFILM
Senior Representative KACTUS
Senior Representative, Weave Bio
Senior Representative, SCIEX
PUREfrex® is a fully reconstituted, cell‑free protein expression system enabling rapid, high‑throughput antibody engineering. It efficiently produces scFv, Fab, and IgG, and supports robust ribosome display (PUREfrexRD) for screening highly diverse libraries. Combined with AI/ML workflows, PUREfrex® accelerates discovery of next‑generation antibodies and cyclic peptide modalities with enhanced precision, flexibility, and scalability for biologics R&D.
Senior Representative, Lonza
Deliverome Bio is a nonprofit research organization mapping surface protein abundance and internalization across the human surfaceome to identify new targets for precision delivery of biologic therapeutics. Doing so first requires a reliable list of surface proteins, yet the five most widely used databases collectively identify 5,546 proteins and agree on only 188. We developed an agentic LLM pipeline that classifies candidates using full-text scientific literature, yielding an open, evidence-linked atlas of the accessible human surfaceome that includes 1,108 candidates absent from all five databases. The atlas and its underlying literature evidence are publicly accessible through our interactive viewer (surfaceome.deliverome.org) and API. We are building on this resource with systematic ORF-based screens of surface protein abundance and internalization, alongside surface proteomics in primary cells and tissues, to enable data-driven discovery and prioritization of receptors for precision delivery.
Senior Representative, Nuclera
Senior Representative, SGS
· Regulatory Adaptation & Global Logistics
· Supply Chain & Node Resilience
· Transitioning & Scaling Operations
Create your personal agenda –check the favourite icon
Senior Representative, ChemPartner
Senior Representative,Minaris Advanced Therapies
Biologics development and manufacturing require the measurement of proteins throughout the process—from monitoring critical process parameters and product quality attributes to detecting impurities and contaminants. However, many protein measurements still rely on laboratory-based assays that take hours or longer to return results, require skilled technicians, and involve multiple reagent addition and washing steps. As a result, important protein measurements are often performed only at selected points in development or as end-point quality tests rather than being used to guide processes in near real time.
Rapid, protein-specific measurements could provide valuable information during both process development and manufacturing. During development, faster measurements could allow scientists to evaluate process changes more quickly and better understand how these changes affect the product. During manufacturing, at-line protein measurements could identify process deviations earlier, potentially reducing lost batches, manufacturing delays, and costs while enabling greater process control and automation. Host cell proteins (HCPs) provide an important example. HCPs can co-purify with a therapeutic protein, and some may affect product stability, efficacy, or patient safety.
Advanced Silicon Group (ASG) is developing NanoSwift™, a silicon-based, label-free electrical biosensor platform designed for rapid, quantitative protein measurements. The technology is intended to measure specific proteins in less than 15 minutes with minimal hands-on time and low sample volume. Because NanoSwift generates an electrical signal and can be multiplexed to measure multiple proteins, it has the potential to support automated, at-line measurements during biologics development and manufacturing.
This presentation will discuss the opportunities created by moving protein measurements closer to the bioprocess, including applications in process development, process monitoring, product characterization, and impurity detection. High-risk HCP measurement will be presented as one example of how rapid, protein-specific measurements could provide information that is difficult to obtain with current methods. Ultimately, faster access to protein data could help biopharmaceutical manufacturers develop processes more efficiently, identify problems earlier, improve process control, and manufacture biologics more consistently.
Conventional biologics are constrained by their reliance on the 20 naturally occurring L-amino acids, limiting achievable pharmacological properties such as protease resistance, immune evasion, and tissue penetration. D-amino acid proteins have long been recognized as intrinsically resistant to proteolytic degradation and immune recognition, but their therapeutic utility has been constrained by the inability to rationally design D-proteins capable of engaging natural biological targets.
We developed a fully integrated, digital and automated platform combining generative AI-based de novo protein design with chemical synthesis to generate "Synteins™"—programmable protein therapeutics built from an expanded set of synthetic building blocks, including D-amino acids. The platform applies folding principles derived from natural proteins to design D-protein binders, inhibitors, agonists, and delivery vehicles de novo against virtually any therapeutic target, circumventing the cell-based production constraints of conventional biologics.
Using this platform, we report the first experimentally confirmed structure of a de novo, AI-designed, all-D-amino-acid protein bound to a natural biological target. This result demonstrates that computationally designed D-proteins can achieve specific, structurally validated engagement with natural targets despite their non-natural chirality. Synteins generated through this approach retain compatibility with human biology while exhibiting enhanced resistance to protease-mediated degradation and reduced immune recognition relative to conventional L-protein biologics.
This work establishes a generalizable platform for designing D- protein therapeutics with tunable pharmacological properties, including extended durability, immune stealth, and expanded tissue access (e.g., brain, kidney, liver, tumor). These findings support the feasibility of SynteinsTM as a new therapeutic modality with potential applications across oncology, immunology, and metabolic disease, and highlight a path toward less frequent dosing and oral delivery of protein-based medicines.
Our team has been developing an interactive machine learning software tool, named Hybrid Interactive Machine Learning Tool to assist physicians in quickly and accurately identifying regions of interest in WSIs by adopting state-of-the-art artificial intelligence algorithms.Integratingsurvival prediction and a conversational AI assistant into HIMLT will make it more clinically useful by helping doctors not just find tumors but understand patient outcomes and use the software more easily. In this presentation, we will talk about how we integrate two modules into the software tool and what to expect fromtheir results.
Bioassays are central to release, stability, and comparability testing of biologics, and must deliver robust, precise results across both GMP and development laboratories. Routine workflows, however, depend on multi-day cell passaging and manual pipetting, thereby introducing cell lot-to-lot and operator-to-operator variability, ergonomic strain, and a heavy documentation burden. This talk presents two complementary strategies for building a leaner, more reliable bioassay operation. Ready-to-use cells replace in-house propagation with a single freeze-thaw step and offer a more consistent starting point, with reduced contamination risk and simpler logistics across sites. Automated liquid handling on Hamilton platforms takes over repetitive, high-precision steps like dilutions and reagent addition, adds consistency, reduces analyst-to-analyst variability, and frees scientists from repetitive manual pipetting. Together, these approaches address both the biological and mechanical sources of bioassay variability, drawing on practical experience implementing both in a regulated environment.
OligodeoxyThymidine (Oligo-dT) affinity chromatography can be used for mRNA purification, exploiting poly-A tail binding to remove impurities. However, process variability and construct-size-dependent behavior challenge yield optimization and GMP consistency. Mechanistic modeling offers data-driven approach to these challenges and accelerate process understanding.
Model calibration achieved R²≥0.95 for UV fit, with <1% parameter error for the target mRNA and <5% for fragments across experiments. Three optimization strategies were evaluated using the calibrated model: (1) Wash Strategy: in silico simulations of GMP wash step matched experimental data, validating predictive fidelity; (2) Dynamic Binding Capacity (DBC) Strategy: simulations revealed steric exclusion of larger constructs from internal pores, with increased load residence time significantly improving binding; (3) Buffer Optimization: screening identified elution conditions that reduced variability and improving pH stability. Model-guided optimization enabled mechanistic understanding of the process, reducing empirical experimentation for wash step evaluation, DBC enhancement, and buffer optimization. This framework accelerates process development, improves product quality, and supports platform-level mRNA manufacturing decisions.
This work was previously presented at the 5th mRNA Process Development & CMC Summit 2026 held in Boston (MA), Sep 30- Oct 1, 2026.
This work was funded by Sanofi.
All authors are or were Sanofi employees and may hold shares or stock options in the company.
Antibody–drug conjugates (ADCs) are powerful targeted therapies but are limited by narrow payload diversity and chemical constraints. In this talk, I will introduce antibody–bottlebrush prodrug conjugates (ABCs) that overcome these barriers. ABCs feature antibodies covalently linked to polymeric bottlebrush prodrugs carrying diverse, tunable payloads at ultra-high drug-to-antibody ratios. This design enables precise tumor targeting, superior efficacy, and minimal toxicity in preclinical models, offering a promising next-generation therapeutic strategy.
·Developed a multispecific TCE platform integrating tumor targeting, T-cell engagement, and costimulation.
·Optimized TCEs enhanced T-cell activation, persistence, and tumor cell killing in preclinical studies.
·Integrated costimulation may improve the potency and durability of TCEs for solid tumors
· Where AI genuinely accelerates discovery and development vs. overpromised applications
· Data quality, integration, and governance challenges across the biologics pipeline
· Regulatory perspectives on AI-driven decision-making in drug developmen
· Single-use vs. stainless steel: capacity planning for unpredictable clinical demand
· Continuous manufacturing and process intensification—readiness for widespread adoption
· Building supply chain resilience post-pandemic: lessons learned and future-proofing strategies