Impressions And Takeaways From BPI West 2025

By Tyler Menichiello, contributing editor

Last week, I took a pilgrimage to beautiful San Diego to cover the BioProcess International (BPI) U.S. West conference. Per the event staff, there were around 800 leaders and industry professionals in attendance, and with so many people to meet and sessions to attend, my only regret was not being able to clone myself to experience more!

However, I attended as many sessions as possible and connected with some great people. I’ll be sharing a handful of exclusive interviews with speakers very soon, but in the meantime, here are some of my impressions and takeaways from this year’s BPI West.

Takeda’s Manufacturability Design Principles, Automating Process Units

The conference kicked off with a powerful keynote presentation by Tracy Ryan, co-founder and chief communications officer at NKore BioTherapeutics. She told the story of how her young daughter, Sophie, was diagnosed with a brain tumor at just eight and a half months old, and the treatment journey that led Tracy to team up with NKore co-founder and renowned researcher Dr. Anahid Jewett from UCLA.

Their collaboration ultimately resulted in the company’s lead allogeneic NK cell therapy, NK101, which Ryan hopes will begin clinical trials in the U.S. in 2026. NKore’s early clinical success reminded me of another company I interviewed last year, NKGen. Though the indications vary, I couldn’t help but think we may see a lot of promise from NK cells as clinical readouts become available in the coming years.

The only other cell-therapy-specific session I made it to was one titled “Addressing the Challenges in Allogenic Cell Therapy Manufacturing” by Takeda’s head of cell therapy process and product development, Amy Shaw. She emphasized the importance and cost efficiency of using small-scale models to optimize large-scale processes and ensure scalability. “The more you can get out of your process, the more doses you can make,” she said. Shaw also touched on Takeda’s manufacturability design principles, which include:

• Scalability to reduce the need for future comparability studies
• Automation (wherever possible) to increase process reliability
• Optimizing cell expansion and CAR expression to improve process yield and maximize doses
• Optimizing fill/finish to minimize the number of cells that need to go into QC, retains, and stability testing
• Minimizing process variability

Some process units that Takeda automates include transduction (to scale-up and include as many cells in the process unit as possible) and depletion steps (getting rid of unwanted cell types or clearance of impurities). “All of these process steps really require refinement of our process losses and just reducing that as much as possible,” Shaw said.

Continued Buzz On Continuous Manufacturing

The conversation around continuous manufacturing isn’t new, but it isn’t going away (at least not anytime soon). I attended the Fireside Chat: “Continuous Vs. Fed-Batch: The Ongoing Debate,” which featured AstraZeneca’s director of bioprocess technologies, engineering, and biopharmaceuticals R&D, Ken Lee, Ph.D., (proponent of continuous) and Jared Dopp, Ph.D., (proponent of fed-batch) an upstream process development scientist at Bristol Myers Squibb (BMS), as well as BPI editor, Josh Abbott. Lee and Dopp discussed some considerations around continuous manufacturing and how their respective companies are incorporating it into their processes.

“Continuous manufacturing is the future,” Lee said early in the conversation, to which the crowd cheered and laughed. He admitted to being biased and explained how AstraZeneca is trying to move away from a platform that is completely fed-batch to a more intensified environment, whether that’s intensified fed-batch, fully end-to-end continuous, or somewhere in the middle.

“For what we’re trying to do, it makes a lot of sense,” Lee said, but the panel agreed: It very much depends on your company’s product portfolio. Some products wouldn’t necessarily benefit from switching over to continuous, Dopp explained, particularly those with high potency (which require lower doses). In such cases, it wouldn’t make much sense to switch over, especially considering the difficulty and costs associated with changing the infrastructure.

The biggest disadvantage right now, Lee said, is really the industry-wide inexperience around continuous manufacturing. The problem is that the people doing continuous now are the first people to do it —especially in downstream, according to Lee. Sanofi and Merck were name-dropped as trailblazers in continuous, but they are keeping that knowledge close to the chest. If the industry wants to see more advances in continuous manufacturing, there will need to be more knowledge-sharing outside of these individual companies and stakeholders, Lee said.

AI Is Great For Automating Tedious Tasks In Biomanufacturing

Unsurprisingly, AI was a hot topic of conversation this year (as I’m sure it will be for the foreseeable future). Not too many sessions had “AI” in the title, but I think most talks were underscored by AI in some way or another. After all, it’s almost synonymous with “technology” at this point.

One session I attended was titled “AI As The Catalyst For Biomanufacturing Excellence.” While I was hoping to hear more about process-specific, technical use cases, I thought the panel did a good job of explaining how most people in the industry are currently using AI — mainly to streamline tedious tasks like writing, data management, and information retrieval.

At that event, one panelist (Reza Farahani, CEO of Katalyze AI) said that AI currently presents the most value for biopharma in operations. Manufacturers can now automate a lot of manual tasks that used to rely on certain expertise, he said. He uses writing deviation reports as an example. Automating such tasks frees experts to focus their time and energy on solving bigger problems. Ultimately, the panelists agreed that your company’s approach to using AI needs to be intentional and that it takes time to use it correctly. On the bright side, it seems we will have to continue keeping humans in the loop (for now).

Optimizing Biomanufacturing With A Factory Digital Twin

Digital twins are digital representations of physical objects, and their use in biomanufacturing is not new. However, rapid technological progress and advancements in AI are resulting in digital twins that are more sophisticated, accurate, and capable than ever.  Joseph Pekny, Ph.D., professor of chemical engineering at Purdue University, gave a talk on using digital twins for process modeling and facility maintenance and design. He attributed the startling rate of progress in the past five years in large part to an increase in computing power.

Pekny outlined the application of digital twins in biomanufacturing as follows:

• Process and Facility Design And Retrofitting
  • Digital twins can both design new facilities and assist in the retrofitting of existing facilities for new products (e.g., where should they go, and which changes are necessary in operations?).
• Alleviating Bottlenecks
  • Digital twins can help identify ways to make a process run faster. Even incremental changes amount to incremental profit, Pekny said.
• Planning and Scheduling
  • In ATMP manufacturing, scheduling is among the more difficult challenges to contend with. Digital twins can provide more accurate predictions and timelines based on real-time data from cell cultures, which is critical for organizing downstream activities.
• Predictive Maintenance
  • According to Pekny, managing production and scheduling maintenance in a way that doesn’t impact production is one of the biggest ways digital twins can save facilities money.

Digital twins tend to become more sophisticated over time, Pekny said. The more people use digital twins, the more the models evolve and mature. This is because every day, model predictions can be measured against what actually happened, and when their predictions are wrong, the models learn and get better. “When the models are very good, then you can control reality to a much higher degree,” he explained.

It’s important to note that purely AI-based digital twins (i.e., one simply trained on past data) are insufficient. The most state-of-the-art digital twins are equation-based models, said Pekny, built on mathematics and the physics of the processes and material balances. That’s not to say that AI won’t dramatically improve the way we use digital twins. “It’s going to have an unbelievable impact on model interpretation,” said Pekny, as well as for model building. However, human judgement will remain invaluable for coming up with the right models.

While digital twins can be invaluable in facility design, nothing beats human expertise and experience. That’s why I’m excited to promote next month’s Pharmaceutical Online Live event, “Facility Design And Validation Considerations For Drug Manufacturers,” featuring bioprocessing experts Herman Bozenhardt and Erich Bozenhardt, as well as facility architect Frederic Grossfeld. We’re going to cover everything from construction sequence to utilities, HVAC systems, and designing for SUS. It’s free to register (which you can do here) thanks to our event sponsor, AES Clean Technology. In the meantime, stay tuned to Bioprocess Online for some more on my time at BPI West in the coming weeks!