There is no inherent conflict between speed and quality. High-quality studies prevent costly setbacks and generate reliable data, ultimately accelerating research programs. A low-quality study is what truly delays timelines by producing unusable or misleading results.

By training on multi-scale data from lab, pilot, and production runs, AI can predict how parameters like mixing and oxygen transfer will change at larger volumes. This enables teams to proactively adjust processes, moving from 'hoping' a process scales to 'knowing' it will.

Raffinose is not a universal solution for glycan engineering. Its ideal use case is for biosimilar matching when you need to specifically increase high mannose content from a low baseline of 1-3% up to a target of 5-8%. Outside this narrow window, it is ineffective or even detrimental and other strategies should be employed.

A structured, three-stage validation protocol can test raffinose in just eight weeks. It progresses from a 96-well plate screen to spin tubes to benchtop bioreactors. Each stage has a clear go/no-go criterion, allowing teams to quickly determine viability for their process without over-investing resources.

Scaling from a T-flask to a bioreactor isn't just increasing volume; it's a fundamental shift in the biological context. Changes in cell density, mass transfer, and mechanical stress rewire cell signaling. Therefore, understanding and respecting the cell's biology must be the primary design input for successful scale-up.

CEO Marc Salzberg clarifies that for their recombinant protein, the difficulty was not in the manufacturing itself but in designing the complex upstream process, purification, and analytics. This innovation became a core asset and "claim to fame," allowing them to transfer a well-defined process to a capable CDMO for scaling.

Many innovative drug designs fail because they are difficult to manufacture. LabGenius's ML platform avoids this by simultaneously optimizing for both biological function (e.g., potency) and "developability." This allows them to explore unconventional molecular designs without hitting a production wall later.

Using raffinose to adjust glycosylation is a regulatory-friendly strategy. Since it is a simple media component adjustment, not an enzyme inhibitor or genetic modification, it aligns with standard process development activities. This avoids intense scrutiny and justification required for more complex methods, simplifying the CMC package.

Titus believes a key area for AI's impact is in bringing a "design for manufacturing" approach to therapeutics. Currently, manufacturability is an afterthought. Integrating it early into the discovery process, using AI to predict toxicity and scalability, can prevent costly rework.