Health-E Law Podcast Ep. 18
AI in Clinical Research: Opportunities, Risks, and the Road Ahead with Bill Kish and Dr. Brad Pruitt of Kenosha AI
Thank you for downloading this transcript.
Listen to the podcast released July 24, 2025, here:
https://www.sheppard.com/multimedia-645
Welcome to Health-e Law, Sheppard Mullin's podcast exploring the fascinating health tech topics and trends of the day. In this episode, Sara Shanti welcomes Bill Kish and Dr. Brad Pruitt of Kenosha AI to explore how artificial intelligence can enhance efficiency and compliance in healthcare research.
About Bill Kish
Bill Kish is the CEO and Co-Founder of Kenosha AI, bringing over 30 years of dynamic experience as a technologist, entrepreneur, and leader across five successful startups. His expertise has led burgeoning companies to flourish into multi-billion-dollar enterprises, solidifying his position as an industry innovator.
A graduate with honors in Computer Engineering from Carnegie Mellon University, Bill’s career has been defined by groundbreaking advancements in AI and machine learning applications. He co-founded Ruckus Wireless, serving as CTO and Board Director, where his contributions shaped the company into a $400M/year business and a leader in the wireless technology industry, culminating in a $1.5 billion acquisition by Brocade.
At Cogniac Corporation, Bill enabled industries to leverage AI-powered visual inspection, serving as the CEO and CTO to drive operational innovation. He also founded Jiggy AI, a boutique AI consulting firm specializing in large language model applications. Additionally, as the organizer of the Silicon Valley Machine Learning Meetup, Bill has fostered a thriving global community of over 10,000 members passionate about machine learning.
About Dr. Brad Pruitt
Dr. Brad Pruitt is the President and Co-Founder of Kenosha AI. With over 25 years of experience in clinical research and healthcare, including 13 years in the Contract Research Organization (CRO) industry, he specializes in revolutionizing clinical trials through advanced AI-powered tools like copilots and GPTs.
Dr. Pruitt is a seasoned executive and entrepreneur with a proven track record of leading ventures to success. He has held executive roles at top-tier CROs, served as the Founding CEO of an acquired startup, and contributed to three successful acquisitions in the past eight years. His prior leadership roles include Chief Medical Officer at Alethium Health Systems, where he developed go-to-market strategies for clinical trial innovation, and Senior Vice President of Medical Affairs at Safe Health, where he drove business expansion into connected diagnostics. In addition to his role with Kenosha AI, Dr. Pruitt is a Principal at Prucor and serves as a mentor and advisor for healthcare and clinical trial technology companies participating in the EvoNexus incubator program.
Dr. Pruitt earned his MD from Michigan State University College of Human Medicine and his MBA from UC San Diego’s Rady School of Management. His academic foundation, combined with his professional achievements, positions him as a visionary leader at the intersection of technology, healthcare, and clinical research.
About Sara Shanti
A partner in the Corporate Practice Group in Sheppard Mullin's Chicago office and co-lead of its Digital Health Team, Sara Shanti’s practice sits at the forefront of healthcare technology by providing practical counsel on novel innovation and complex data privacy matters. Using her medical research background and HHS experience, Sara advises providers, payors, start-ups, technology companies, and their investors and stakeholders on digital healthcare and regulatory compliance matters, including artificial intelligence (AI), augmented and virtual reality (AR/VR), gamification, implantable and wearable devices, and telehealth.
At the cutting edge of advising on "data as an asset" programming, Sara's practice supports investment in innovation and access to care initiatives, including mergers and acquisitions involving crucial, high-stakes and sensitive data, medical and wellness devices, and web-based applications and care.
Transcript:
Sara Shanti:
Welcome to Health-e Law. I'm Sara Shanti, a healthcare partner with Sheppard Mullin, and I have the pleasure of hosting today's episode. We do know that more than ever there's a need for efficiency and improved resources in the clinical research space. On that note, I'm thrilled to be joined today by Bill Kish, who is the CEO and co-founder of Kenosha AI and Dr. Brad Pruitt, the president and co-founder of Kenosha AI, who are specifically focused in this area of clinical research and regulatory intelligence and efficiencies. And they're going to talk us through a little bit today about what their company is doing to close the gap on the need for resources to keep the momentum of clinical research. Bill and Dr. Pruitt, lovely to have you with us. Thank you for joining us today.
Bill Kish:
Thanks. Super glad to be here.
Sara Shanti:
Wonderful. So let's get started right away. I know that we're seeing AI as a buzzword in kind of everything that we do and kind of every page that we turn these days in the news. So maybe you could tell us a little bit about Kenosha AI and how you see it playing a real role in healthcare and research operations.
Dr. Brad Pruitt:
Excellent, thank you. Our focus really is on operational use cases for AI in the clinical research space. And there's a couple unique challenges that this industry faces, whether that's on the sponsor side and the CRO side is that there are a lot of global regulations that continue to evolve weekly, sometimes even faster than that. They have to be followed and you have to stay compliant to those global regulations, FDA, EMA, ICH, et cetera. And as well, you've got a lot of internal SOPs that you have to follow, protocols that you have to follow in actually operating the clinical studies.
So there's a lot of evolving instructions and regulations that have to be followed. Most of these organizations, especially as they grow and become larger organizations, have a ton of knowledge trapped in silos and SharePoint folders and making it really difficult for especially new folks that joined, don't know where all the data is buried, very limiting to get cross-organizational access to all that information. A lot of standardized reports and documents that are getting created that is very monotonous and very standardized and ripe for automation. And just along with that, a lot of repetitive tasks and workflow in the industry that very ripe for a lot of the advantages that AI has in its reading and writing and reasoning.
Sara Shanti:
Thank you, Brad. And Bill, maybe you can expand a little bit on that for how the day-to-day patient, especially those patients with rare disease and who are looking to get involved in clinical studies, how that ultimately means something positive for patients.
Bill Kish:
AI has an incredible potential to bring efficiencies to these processes. Everyone expects AI to have an impact on medicine in many ways and on drug development. We're already starting to see research applying AI to customize therapeutics, things of that nature. And so the backlog of demand for clinical trials is expected to grow. So I think really the only path forward is using AI to increase the efficiency and increase the quality of the data coming off of these trials.
Sara Shanti:
When you take one of Kenosha AI's incredible products, which I'd like you to explain a little bit further of what they can do and kind of some of your recent rollouts on your AI chat or your regulatory chat, who is the user of that product in which they're going into the lab, they're going into their health system dedicated to a massive research study or a funded resource study with a limited budget. How do you kind of pose these products as having immediate and real effect on their mission?
Bill Kish:
Yes, I think one of the real abilities of this AI technology now, it has this superhuman skill in being able to read, that current AI models can read a hundred pages in a second, in some cases 200 or more pages of information in a second, and really synthesize that into responses. And so there's a whole class of tasks that would've basically not have been done before that the AI can now do in seconds. Being able to research topics and regulatory intelligence, being able to work through your entire SharePoint site and extract some piece of information that was hidden in some document that everyone forgot about, these are the real superpowers that are bringing efficiencies to the operations right now.
Sara Shanti:
Brad, looking forward then, once you use an incredible resource and it's time-saving, it's accurate, you have trust in it, how to work it well, how do you see this momentum building and where's clinical research going because of the AI power that you've been able to apply?
Dr. Brad Pruitt:
I think one of the big aspects of the amount of data coming in to the trials from all the extended data points as the trials are getting smaller in average size, but the data points are continuing to get more complex and growing in the continuous Fitbits and connected devices that are gathering data to just the abundance of data that's coming in, needing to have that real-time ability to monitor all of the inflow of data, looking for anomalies, and to be able to raise a red flag where it used to be you didn't see that until you did an interim analysis or delayed look at the data to be able to see any signals. And some of those could be safety signals that they need to address sooner than later, and in real time is fantastic.
Sara Shanti:
Bill, I know you have such deep experience in the venture space, in AI and machine learning, software development and product development, and so I'd love to hear your thoughts on what's the risk and what's the concern around hallucinations, synthetic data, that trust level in the AI deliverable?
Bill Kish:
Sure. That's a very interesting question that a lot of people are pondering right now. The AI models, they're really amazing in their capability, and interacting with them, you're often amazed at their breadth of knowledge, but the key thing to understand about in the AI model like GPT-4 or the newer variants of that, they don't have a perfect database of all their information. At training time, they're basically reading all the content of the internet and they're distilling that. They're building a kind of statistical model. It's one of the reasons why they work so well with language is they have a very rich statistical model of language. They learn a lot of facts, but any single fact they're not necessarily able to retain. They have this general statistical understanding of many things, but individual facts are not necessarily recalled when you ask them a question. And so that's where the hallucinations come from.
They imagine things that might ought to be but aren't necessarily true. And so the way to fix that is to actually just give them material to read and instruct them to answer not from their memory, but from what they're reading in front of them. And so then the goal is when interacting with a person asking questions, the system has to do a search on the database, kind of like a Google search, but using a little bit of different techniques and typically using a curated set of knowledge, not answering from the open internet, but answering from a set of vetted documents. So for example, if you're asking regulatory questions, you'd want to refer to the actual material from say, fda.gov or whatever country's regulatory agency you're interested in.
So this technique of taking a curated set of documents and allowing the model to read them, and as I mentioned, the models can read tens, hundreds of pages in just a few seconds and extract the relevant information from that and respond factually with that information. This is the single most effective technique for addressing hallucinations is to make sure the models have the right context in front of them and are instructed to answer based on that. It's not foolproof so you still need a sharp set of eyes and references and based on the risk associated with what you're doing, you need to have a proportionate level of verification for the information that the AI model's giving you.
Sara Shanti:
We see a lot of excitement over AI and some of these new tools as well as a lot of hesitancy. So should clinicians, researchers, where should healthcare really feel ready to jump in today versus kind of waiting or being patient and seeing where there is some development that they then feel more comfortable in jumping in? Brad, maybe you can chime in too, but I'd love to hear, is this the moment to take advantage? Or for those that have some hesitancies, is it a good consideration to wait a little bit and to see where some of this development goes?
Bill Kish:
What I've seen with AI models, if you're interacting with a very capable AI model like ChatGPT, it's sometimes magical in the results that it can give you, but it's really not magic under the covers. It's really not magically good at all tasks that you might want it to do. And there's a certain bit of skill involved in building intuition around what tasks it actually can do and how to actually instruct it to get the outcomes that you want. It's sometimes like managing a person, finding the right person for the job, knowing if someone's going to be able to, capable of doing the job that you want, and understanding how to explain to them what you expect, these are all skills that develop with your use of the AI models. And by beginning to interact with these AI models, you build this intuition around how to best make use of them, when not to use them, and how to refine your style of interacting with them.
So I think for quite a long time, we're going to be working with these models that still need to be told what to do and interact with these AI models. So the short answer is, it's better to get these skills to begin developing these skills and how you use AI. And just like any tool, there's significant differences in how a master craftsman can yield the tool versus a newcomer. And so I really do think that AI right now is looking more like some tool that really can absorb a lot of craftsmanship and how you use it. So yeah, my recommendation would be to get started, even if it's just experimenting with it, building that intuition around how best to use it.
Sara Shanti:
That's excellent. Well, I like the way you described that. It's a craftsman, you have to kind of learn how to make that tool work for you. And that really jives with a lot of what we hear from our clients and from the industry, and we just have the pleasure of serving such a wide swath of the industry that we're really hearing that using it as a tool, not a replacement. It can't practice medicine but using it as a tool can really make everyone sharper and better and ultimately hopefully save lives or save a community from a potential pandemic. Bill And Brad, I want to allow you to both step up onto the soapbox for a moment and either say who should reach out to you or a takeaway to our audience so that they can really leave with that great last nugget from your expertise?
Bill Kish:
I would say when thinking about things that you can use AI for, really focus on those superhuman tasks that play to its strengths. Right now that really is reading hundreds of pages and working through updating of your SOPs when there's a change from E6(R2) to (R3), something that would involve otherwise reading and editing thousands of pages that AI makes real quick work of that type of stuff. And its stuff that no person really isn't either an expert in or really wants to do. And so those types of tasks are really where the nuggets are right now with this kind of first wave of AI.
Sara Shanti:
Excellent. And Brad, anything you want to add to that?
Dr. Brad Pruitt:
Yeah, that's a big push right now as we've really gotten deep into the use in regulatory intelligence, AI for regulatory intelligence and looking at the harmonization updating of SOPs. But as well as document writing, standardized documents that have some sort of a template and instruction around them of how to be written, those can definitely be automated and reduce a lot of monotony and folks' work how to have those generated. At least get those 80, 90% of the way there and have the experts finish up that last 20% in the last mile to get it proofread and edited.
I think those are great use cases of where you can take care of 80% of the work with AI and then let your SMEs take care of the last 20% of where it's really important to have number one, eyes on, but have the right folks that are interpreting it and translating that correctly for the given use case of whether it's a patient informed consent or whether that's a clinical study report, that it has the right interpretation, you definitely want to have the human eyes on your critical information before that goes out and gets a final stamp of completion and approval.
Sara Shanti:
Well, thank you for your commitment in pushing clinical research forward and supporting our clinicians and the research out there. We're really excited about both your RegChat and your compliance AI intelligence. For our audience out there and for folks that are going to take a look at this episode, who do you want to make sure knows that they should certainly check out the show notes, check out your products, and I believe you even have a free trial available?
Dr. Brad Pruitt:
Yeah. For folks that want to go check out our RegChat. We have a free trial available at regchat.com. You can go and sign up for a free account and get a taste of that on a test drive. Our main customers are the sponsors running, operating the clinical studies or the CROs that they have contracted to manage those clinical studies. So again, we're really focused on clinical trial operations in the life science industry, from pharma, biotech, med device, and the CROs that are operating those studies.
Sara Shanti:
Excellent. And thank you for joining us today, Bill and Brad.
Bill Kish:
Thank you.
Dr. Brad Pruitt:
Thank you.
Contact Info:
Additional Resources:
Kenosha AI - Kenosha AI is currently offering a free trial of its RegChatTM, which is an AI-powered Clinical Regulatory Guidance Assistant that provides a simple chat interface for answering questions about global regulatory guidance using AI and official regulatory guidance documents with referenced summarizations and multi-agency comparisons. Find it at RegChat.com.
* * *
Thank you for listening! Don't forget to SUBSCRIBE to the show to receive new episodes delivered straight to your podcast player every month.
If you enjoyed this episode, please help us get the word out about this podcast. Rate and Review this show on Apple Podcasts, Amazon Music, or Spotify. It helps other listeners find this show.
This podcast is for informational and educational purposes only. It is not to be construed as legal advice specific to your circumstances. If you need help with any legal matter, be sure to consult with an attorney regarding your specific needs.