What if the answer to battling antibiotic-resistant infections isn’t a new antibiotic, but harnessing viruses that have been quietly dominating bacterial populations?

Bacteriophages, viruses that target and kill bacteria, have been saving lives for a century, but their true potential is only now being unlocked by modern machine learning. The race isn’t just about discovering effective phages; it’s about deploying the right therapy, personalized to the patient, before time runs out.

This is part two of the conversation with José Luis Bila, CEO of Precise Health, a personalized phage platform aiming to improve and treat bacterial infections employing a combination of bacteriophage technology with data science and machine learning. 

Key Topics Discussed

• Continuous bacterial–phage evolution makes full isolation impossible; need for adaptive and in silico–engineered solutions.
• Overview of bioreactor-based production, timelines, GMP requirements, and quality control for reliable output.
• Combining CDMOs for standard phages with in-hospital production for specialized cases; disparities in global infrastructure and LMIC challenges.
• Exploration of mobile or miniaturized bioreactors, downstream processing bottlenecks, and affordability pathways for LMICs.
• Purification, shelf-life, and formulation issues compared to monoclonal antibody or viral vector production.
• Dual regulatory challenges for phage therapeutics and AI-driven diagnostics; regional flexibility and evolving approval pathways.
• Transition from AI-matched to AI-designed phages; commitment to global accessibility and a call to action for innovation against AMR.

Episode Highlights

• The challenge of evolving bacteria and phages, and the question of whether it’s possible to keep up with nature’s pace through engineering new phages in silico [00:00]
• Overview of bacteriophage production: complexity, types of bioreactors used, and comparison with chemical synthesis [02:54]
• Bioproduction logistics: using CDMOs vs. in-house hospital production and the real-world timescales for manufacturing [02:32]
• Barriers for smaller or less funded hospitals, especially in low- and middle-income countries, with thoughts on hospital infrastructure differences worldwide [04:41]
• Creative solutions for cost-effective phage production in remote and underserved regions, such as the potential for single-use or mobile bioreactors [06:01]
• Why downstream processing and ensuring product purity is a bottleneck; the need for miniaturization and economic scalability [06:40]
• Parallels and differences in downstream processing between bacteriophages and viral vectors [09:15]
• The vital role of stability and shelf life for phage therapy logistics and economic viability [09:15]
• Regulatory pathways for phage therapy in Europe and beyond, including magistral preparations, ethical approvals, and adapting to digital tools for selection [12:39]
• The future vision: routine clinic entry through matching existing phage libraries, with longer-term goals of engineering bespoke phages via AI when necessary [15:52]
• Jose’s perspective on building global infrastructure and making phage therapy cost-effective and universally accessible [18:08]
• A call to action for scientists and innovators to pursue real-world solutions for antimicrobial resistance without delay [19:11]

In Their Words

Can you isolate all the bacteriophages in the world? No. Even if you do isolate them, the bacteria are evolving and the bacteriophages are also evolving. How do you catch up to those evolutions altogether? It's insane, right? We start with the matching and maybe putting the bacteriophages into practice, etc. But in cases where you do not find any bacteriophage or there is no antibiotic that can be used, then what if we could engineer new bacteriophages in silico that could then be taken into production?

Episode Transcript: Fighting Antimicrobial Resistance: How AI Cuts Phage Therapy Access from 6 Months to 5 Days with José Luis Bila - Part 2

David Brühlmann [00:00:33]:
Welcome back to the Smart Biotech Scientist. I'm David Brühlmann, your host, and this is part two with José Luis Bila, who is the CEO of Precise Health, where we're exploring the business and regulatory sides of AI-powered phage therapy. In part one, we heard his compelling personal story and learned how artificial intelligence can revolutionize treatment selection against bacterial infections.

Now we're diving into the nitty-gritty: production scalability, regulatory hurdles, and the economic realities of bringing personalized phage treatments to market. Discover how AI is making life-saving therapies both accessible and affordable for patients worldwide.

How do you produce these bacteriophages, and also, how long would it take, for instance, in a case where we have 80 or 90% coverage, but unfortunately, a patient comes in with that 10% we have not covered? How fast could you start the bioreactor, and how long will it take? Also, can you tell us a bit about what kind of bioreactors we're using?

José Luis Bila [00:02:54]:
Producing bacteriophages is actually quite complex by itself, but to be honest, it's not the most complex thing—it’s been done for ages. I'm not a bioprocessing person, but I would reckon that it's a little less complex than synthesizing a chemical molecule. So, I’d say it’s quite manageable.

The type of bioreactors we use are standard bioreactors—the same ones used for bacterial growth are the same ones used for bacteriophage growth and production. In terms of timing, it depends on who is doing the production. Some hospitals already have their own in-house capabilities, which is a major advantage. Most hospitals, however, don’t have that, so you need to go to a commercial CDMO. For commercial CDMOs, the minimum timeline is typically two to four weeks, assuming they have an empty bioreactor and personnel ready to start production.

Then there’s another issue: stability and quality control. You need to go over all of these steps carefully. Everything needs to be considered—essentially, it’s the classic GMP process. All the quality management systems need to be in place to ensure production is efficient and correct.

David Brühlmann [00:04:22]:
I guess a dual system could work very well. You could use a CDMO to produce the 80–90% of bacteriophages you need regularly, and then have smaller bioreactors in a few hospitals to quickly produce whatever is lacking. That could be a solution.

José Luis Bila [00:04:41]:
Yes, that’s correct. The reality today is that only one hospital in Switzerland that I know of has GMP production capabilities. Most hospitals in the world do not have GMP capabilities. However, when they need phages for patients, they have processes in place that are not fully GMP but could be considered GMP-like, with proper sterilization methods, etc., to deliver bacteriophages to patients. Some hospitals have already invested in this capability.

But it’s a big assumption that most hospitals worldwide will have this. In Europe, many large, university hospitals may have this capability, but regional hospitals? That’s much more difficult. In US, again, in big hospitals, bigger cities and well-funded hospitals might manage, but rural areas? Not so easily.

We also have to consider that most antimicrobial resistance cases are coming from LMICs—low- and middle-income countries. Whatever strategy we develop today, we must account for that, because that’s where the most impact can be made.

David Brühlmann [00:06:01]:
Could we envision some mobile facilities somewhere in a remote area of Africa or Southeast Asia? Because as you mentioned, that's where the most cases will be. I think in the Western world we could cover it with a few centers and then quickly ship it if needed. But where it will take time is especially in those remote areas and probably that's also where the expertise is still lacking today. How could we go about producing that in a very economic way, in a cheap way, maybe with single use technology or some mobile reactors? We probably need some creativity there.

José Luis Bila [00:06:40]:
Yeah, exactly. I think if we manage to build a single use bioreactor for each patient that is cost effective some way, that will be a billion dollar idea for sure. Because this is something that I think all personalized approaches would need somehow. And the issue I feel is not really on the bioreactor itself, it's really on the downstream processing of things. Because you want to be able to build a bioreactor but you need to combine it with the downstream processing in a way that you get something that is pure. But how do you miniaturize it in a way that is going to be cost effective? If anyone has some ideas, we are happy to collaborate on those things because yeah, this is a billion
dollar idea for sure.

But for the LMICs, the way I see it is that in Switzerland, we are fortunate enough to have these big hospitals that make us think outside the box, initially without the thinking of the costs or something that we can already kind of start implementing, thinking of ideas of how to make it work. And then in a second phase we will have to be able to decrease the cost so that we can feed everybody else who does not have the same economic power.

Now for LMICs, we will have to implement it here first, make sure that it works and then how do we go down the line. That's going to take a bit of time. But I am from LMIC myself and like I said my parents died from the antimicrobial resistance. And in that case, even if we had bacteriophages today that could have saved them, I wouldn't have benefited them by just us focusing on financially powerful markets. So I need also to think of that and this is something that attracts me to my team every day. Whatever you do think long term, three to five years from now, if you were in Uganda, for example, how would this be used and how would it be implemented in a hospital where does not have the same capability as one big hospital in Switzerland?

David Brühlmann [00:08:47]:
Yes, that's a very good point, Jose. Start with the end in mind because finally it's about the end user and how are we going to bring it to them. Even if we have the best technology, but we cannot deliver it or it's too expensive, it will be of no use. Now just I want to double down a bit on the process again. So since you mentioned the downstream, can you tell us how different the downstream is from a standard, let's say MAP process or like a viral vector process? Is that very similar or is it very different?

José Luis Bila [00:09:15]:
From what I understand, it’s quite similar. I’m not doing this day-to-day, and I’m not very familiar with viral vector bioprocessing, but phages are viruses themselves, so I wouldn’t expect anything drastically different.

One thing I want to highlight, which I mentioned earlier, is stability—this is extremely important. Let me step back a bit: some doctors are concerned with logistics, like how to get the right phage at the right time. That’s one issue. The second issue is the lack of systematic clinical data. We have enough treated cases to show promise, but they haven’t followed classic clinical trial protocols with standardized data points that can be consistently implemented.

This is one issue. On top of that, there have been some clinical trials done back in 2017, if I’m not mistaken, on phage therapy. In one trial, they used a cocktail of phages at 10^6 PFUs for wound infections. Over time, they noticed that stability was not great. Instead of administering 10^6 PFUs to a patient, they ended up giving only 10^2 PFUs per milliliter. This is clearly insufficient—so instability was a major problem.

Another issue in that specific trial was that each patient’s bacterial isolate needed to be tested against the phage cocktail before administration. You couldn’t simply assume that the cocktail would work for every patient, unlike antibiotics, which are more broadly effective.

Returning to the stability issue: if we want to implement our strategy—taking the bacterial fingerprint of a specific region and stocking phages—we need to keep them stable for a long time. That’s crucial for the economics. If a batch only lasts three months, how likely is it that a patient will need that batch within that timeframe? If we can increase shelf life to two or three years, the economics make much more sense. We can produce batches, store them, and eventually someone will use them.

This is also where miniaturizing the bioreactor—your idea of a single-batch system delivered directly to the patient—comes in. Until we achieve that, we need proper phage stability. One way to achieve this is through lyophilization, which various groups are actively exploring to increase stability.

David Brühlmann [00:12:17]:
That's a very important point, José. I just want to also briefly touch upon the regulatory aspect because finally it's a drug you going to be used to treat patients. How do you navigate this with the health authorities? Especially as you're using machine learning and AI approaches to identify the fate. Is that relevant or not? Or how do you go about that?

José Luis Bila [00:12:39]:
There are two regulatory fronts, and this is always a big question, no matter which investor we talk to: What are the regulations around this?

The first front is phage therapy itself. If I’m a doctor and want to administer phage therapy, which set of regulations do I follow? The second front is the process to identify the right phages and produce them before they can be used under the first regulatory framework.

For the first front, many countries—and many stakeholders in the phage world—are exploring ways to implement this in hospitals. For example, in Belgium, they use magistral preparations for each patient, so GMP is waived. Hospitals prepare phages in-house and administer them directly. They maintain a library of bacteriophages, and doctors follow a manual process to identify and produce the required phage.

Last year, around November 2024, Portugal announced that it would follow Belgium’s model: bacteriophages can be used as long as there is a clear patient need, no alternative exists, and ethical approval is granted by the hospital committee. Some other European countries and even stakeholders outside Europe are following similar approaches. So there is some regulatory relaxation. However, the European Union still requires more clinical trials to establish certainty.

The second front is the use of machine learning for phage identification. This is still not fully mature for direct regulatory acceptance, but there are precedents. Consider antibiograms in hospitals: the process is the same, except instead of testing antibiotics, we test phages. If digital antibiograms exist, why not a digital phagogram?

We are not building a platform to be used recklessly. We want intelligent guidance for doctors. We are also pursuing CE marking and all necessary regulatory approvals. This builds confidence for medical doctors to use the platform and ensures we follow all regulatory pathways for clinical applications.

David Brühlmann [00:15:47]:
What is your vision for the future of this machine learning powered phage therapy?

José Luis Bila [00:15:52]:
It’s a deep question. There are many fronts to consider. Today, the most natural entry point for clinics in Europe—or anywhere that isn’t yet using phages—is natural phages. You extract them, isolate them from multiple sources, and then identify the right ones. Managing this identification and reducing logistical challenges is where machine learning can help first.

Looking ahead three to five, maybe even ten years, we believe this is where we can have the most impact. Consider this: estimates suggest there are 10 to 30 million bacteriophages on Earth—a huge number. Can you isolate them all? No. Even if you could, bacteria and phages are constantly evolving. How do you keep up with these changes?

That’s why we start with matching phages from existing libraries. But in cases where no suitable phage exists or antibiotics fail, what if we could engineer new phages in silico and take them to production? This is the vision of Precise Health: a system where you first match from existing libraries, and if nothing fits, you can engineer new phages with a single click, produce them, and potentially use them for the patient.

Of course, this is far more complex than it sounds and will take time. Implementing engineered phages is harder and slower, which is why we are starting with the matching approach first.

David Brühlmann [00:17:54]:
Wow. The future definitely looks bright or interesting at the least.

José Luis Bila [00:17:59]:
Yeah, true.

David Brühlmann [00:18:00]:
This has been great. José, before we wrap up, what burning question haven't I asked that you're eager to share with our biotech community?

José Luis Bila [00:18:08]:
Burning question? It’s not really a question—it’s more of a vision for me. Bacteriophages have existed for 100 years, and we know they work. They’ve been used multiple times successfully.

The real challenge is building the infrastructure and foundation to implement phages cost-effectively and accessibly, so they can have a global impact—not just in Europe, but worldwide. We believe machine learning is key to achieving this, and this is what we’re trying to demonstrate to the world.

So, maybe not a burning question, but definitely a burning comment.

David Brühlmann [00:18:59]:
Fantastic. With all that, what we've covered today, what is the number one takeaway you want our listeners to walk away with?

José Luis Bila [00:19:11]:
We don’t need to wait to find solutions for people who are dying. This could affect anyone—it could be us, or your loved ones. Antimicrobial resistance is one of those urgent issues, and current systems aren’t working.

If you have an idea to help—even if it’s outside your usual field—go out and try it. Eventually, someone will find the right solution. I think that’s the key takeaway.

To all the listeners out there, thank you so much for taking the time to listen.

David Brühlmann [00:19:54]:
Thank you so much for moving the needle, José . And thank you so much for investing your time and passion to solve a real problem of humanity. Also thank you very much for coming on the show. It has been a huge preasure to talk with you. Before we go, where can people connect with you?

José Luis Bila [00:20:11]:
We can be found at www.precisehealth.io, that's our website. But for those who want to reach me directly, it's jose.bila@precisehealth.io . I'm happy to connect with anyone who might have questions or I can help with anything.

David Brühlmann [00:20:30]:
Fantastic. I will leave the links in the show note so reach out to José . And José , once again thank you very much much for coming on the show. It has been a huge pleasure.

José Luis Bila [00:20:40]:
Thank you so much. It was a pleasure to me. My first podcast. I hope it's impactful to anyone who is
listening. Thank you so much.

David Brühlmann [00:20:47]:
José's vision for AI powered phage therapy offers real hope in our fight against antibiotic resistance. His journey from personal loss to breakthrough innovation proved that biotech can truly save lives. Do you have questions about process development or manufacturing challenges? Well, book a free consultation with me at www.bruehlmann-consulting.com/call and I'm happy to help you get started. And please rate us on Apple Podcasts or your favorite platform because every review helps us reach more biotech scientists ready to make their mark. Thank you so much for tuning in today and I'll see you next time.

All right, smart scientists, that's all for today on the Smart Biotech Scientist Podcast. Thank you for tuning in and joining us on your journey to bioprocess mastery. If you enjoyed this episode, please leave a review on Apple Podcasts or your favorite podcast platform. By doing so, we can empower more scientists like you. For additional bioprocessing tips, visit us at www.bruehlmann-consulting.com. Stay tuned for more inspiring biotech insights in our next episode. Until then, let's continue to smarten up Biotech.

Disclaimer: This transcript was generated with the assistance of artificial intelligence. While efforts have been made to ensure accuracy, it may contain errors, omissions, or misinterpretations. The text has been lightly edited and optimized for readability and flow. Please do not rely on it as a verbatim record.

Next Step

Book a free consultation to help you get started on any questions you may have about bioprocess development: https://bruehlmann-consulting.com/call

🧬 De-risk CMC development and get decision-making guidance with a new AI platform that transforms CMC overwhelm into predictable development success (launching early 2026). Join the waitlist here: https://david-jkhjdoje.scoreapp.com

About José Luis Bila

Dr. José Luis Bila is the Co-founder and CEO of Precise Health, a company dedicated to making phage therapy faster, smarter, and more accessible through AI-driven innovation. He earned his PhD in Chemistry from EPFL and began his career in life sciences consulting, advising global biotech and pharmaceutical firms on strategy and innovation. He later joined a MedTech startup developing rapid STI diagnostics.

Blending scientific rigor with entrepreneurial vision, José leads Precise Health’s strategy, product development, and partnerships. His personal experience—losing both parents to antibiotic-resistant infections—fuels his mission to bring effective, precision therapies to patients where traditional antibiotics no longer work.

Connect with José Luis Bila on LinkedIn.

He is also a biotech technology innovation coach, technology transfer leader, and host of the Smart Biotech Scientist podcast—the go-to podcast for biotech scientists who want to master biopharma CMC development and biomanufacturing.  

Hear It From The Horse’s Mouth

Want to listen to the full interview? Go to Smart Biotech Scientist Podcast. 

Want to hear more? Do visit the podcast page and check out other episodes. 
Do you wish to simplify your biologics drug development project? Contact Us