Genflare AI Advisory — Patient Access Automation

Hello.

Hello.

How's it going?

Good, how are you?

I'm good. I am good. I'm Allison Herald. And we will wait a few minutes because we usually start five after.

I like that it flips it a little bit, but I think it's probably better. I've never had a policy like that, but I'm guessing it's better.

Not so much a policy, it's just a practice that we've integrated because you're usually running over one meeting and you're trying to get to the next one. It gives people a little opportunity to make it to the next meeting without rushing. So we're looking forward to hearing all about — is it Genflare? Okay. And you met Patrick at Asembia. And is it just you from your group or is somebody else joining?

Greg Johnsen, my co-founder, will be joining. He should be along shortly.

Okay, perfect.

Are you in New York?

No, I'm actually in LA.

Okay, got it. I saw the New York.

I am from the east coast — I moved here from Philly. Although my family's from New York.

So you're not an expat then, is what I'm hearing?

Oh, I wouldn't say that. I moved out here 10 years ago, so at some point I'll get back to the east coast. I see Greg has joined and there's Patrick — hey Patrick, how's it going? Patrick, you got a different hat on, don't you? I love New York, I would live there if I could afford to, although LA is not much better.

Are you in the Thousand Oaks area where Amgen is, or some other part?

No, I'm actually in the Valley. I was taking a public health class and I think Singapore and New York were tied for the most expensive places to live in the world. Are you in New York?

No, no, we're NorCal folks. I'm from Sacramento and live in Sacramento now because I just moved back here, but was in the Bay Area for 20 years. I did school in San Diego and had lots of friends in LA, so — California boy, born and raised.

And Greg — I'm in Berkeley. Oh, okay, so you're both in California. So this worked out time-wise for everybody. Excellent.

It's rare that we have pharma client meetings that are west coast, so it's nice. Sometimes it's 7 AM or even 4 AM, you know.

Can be rough. Hey Jess, I think we're waiting on John, right? Okay, let me just look — I was just on a call with him not too long ago. He's probably getting situated. [Crosstalk about calendar accepts and meeting logistics.] Should I ping him? Yeah, let me ping him. Let's just see — another minute. I think the gang's all here. That's okay. We were just ribbing you about being late, John.

Let me do a quick intro and background and then we can do introductions just around what each of our roles are. I think I've spoken to each of you individually around different vendors that I had the opportunity to meet with at Asembia. And Genflare was one of the ones that I think piqued my interest the most, just given exactly where we are right now in our technology roadmap.

So if you guys remember, Genflare was the vendor I talked about that is using AI coding to help accelerate CRM development — custom CRM integrations, both inside of Salesforce, on top of Salesforce, and just in general looking at how do we accelerate time to deployment. I think I told you guys about a scenario in which they participated and helped run an RFP and develop software by the end of the RFP, which was an interesting use case and demonstration of how much we can actually speed up development if we're utilizing some of the new AI tools in the ecosystem. So that's them in a nutshell, but I'm sure they're going to do much more justice to what I just wanted to overview. You guys know me — right now I work as one of the marketing leads for one of our rare disease brands, which is Krystexxa. I also work on Amgen By Your Side kind of above-brand marketing, but work really closely with Jessica and John on our overarching technology approach to the market, how we represent ourselves. But these two kind of run the operations and the technology roadmap. Jess, you want to start?

[Jessica] So, part of our RDBU patient access team as Patrick mentioned. Somewhat recently taking over our strategy and technology roadmap and trying to move ourselves from what we have today, which is antiquated processes and technology, to something that's a little bit closer to this decade and figuring out the right approach there. We've got the larger initiative from a one-CRM perspective of consolidation across our USPSPs, but then we've got some specific needs that are a little bit more unique to a rare disease perspective that we need to address just because of the high-touch model and the high-touch support that we provide, which is obviously different from a gen med portfolio or something like that. So working closely with this group of folks as we're sorting that out and charting our path forward.

[John Bano] Hey guys, John Bano. I'm the hub services lead for RDBU. So that is our outsourced BV, copay, and PAP programs and our insourced case management team. I think Jess said it very well — that we're right now sort of solidifying the foundation from an overall service offering, so that as we move forward we have the opportunity to accelerate from what is currently not this decade and start to move at pace. Very nice to meet you guys. I also have to say, Patrick told me the story that you guys shared and I told that story to two people already. So I was very excited about meeting with you guys.

We'll try not to disappoint you.

[Allison] I'm Allison Herald and basically what Jess said — with a specific remit around looking at digital capabilities. Nice to meet you all.

Well, I'm Bradford Lee. Greg Johnsen is my co-founder. My background is biotech, startup, and product management — for Abbott when Abbott was still part of AbbVie. So I come from this world but have had the entrepreneurial itch my entire career. Spent the last 28 years building and commercializing digital health solutions for the healthcare industry — half that time providers, half that time pharma, all consumer-facing engagement. Patients, providers, nurses, what have you. The last company that Greg and I built together was actually one of the first agentic platforms for patient engagement for patient services. The company's doing well — new leadership, new funding since we left about a year or so ago. But we have now founded an AI advisory and services company for commercial organizations at pharma. We'll get into the specifics of what we do, but the way to think of us is sort of the new-age services and strategy partner for the era, and we can talk about what that means and why it's different. Part of the work that we do is to accelerate the building of custom, bespoke enterprise applications for pharma. So I have a deck, I have the PA intake RFP case study — so we can go through all that. But excited to dig in with you all. Patrick, thanks for arranging this meeting.

Yeah, of course.

[Greg] I'm Greg Johnsen. About an hour and a half away from Bradford — he used to live in the Bay Area and then moved up to Folsom where he grew up, couple little kids, a little easier to live up there. That left me down in Berkeley/Oakland where it's a little more urbanesque. I've been in the valley here for about 30 years. I've started a few different companies. I'm mostly a technology person that ran product organizations, sales organizations, marketing teams — did a little bit of everything. And the last decade spent in life sciences, healthcare, specifically around patient engagement and the AI platform that we built with Fluence, which used to be called LifeLink. Happy to meet you, looking forward to this session. This is our favorite topic on the planet — AI and what it can do.

Quick question for you both — so would you consider yourselves a startup? And are you currently working with other pharma now?

Yes, startup. Working with other pharma and medtech companies — run the gamut from large pharma like yourselves to mid-tier, sort of growth-stage if you will. And all of our network and experience over the last 10 years or so has all been in specialty pharma.

Cool. Well with that, I might just jump into the content unless there's anything else anyone wants to add. I move quickly — somewhere between 15 and 18 minutes of prepared material and then I think we just open it up for discussion and questions and kind of see where this goes, if that sounds like a good plan.

Perfect.

So. We all know AI is a seismic moment for pharma commercialization — we know that it's a top strategic priority for every pharma company on the planet, we know tons of money is being invested in it. But the reality is less than 10% have scaled a single AI use case to production. This comes from a Deloitte 2026 life sciences study, January 2026. And we believe that progress is slow not because of the tools but because it's about reinventing the way that you work.

And our vision for what the commercial team of the future in pharma looks like is very different than what a commercial team does today or how it does it today. We see this across five dimensions: your strategy — how to use AI as a strategic thought partner to get better strategy but also to accelerate execution. We think the way that you structure yourselves in terms of tech and people and process is different — we don't believe that technology will be buried in IT, held by outside vendors primarily, but actually embedded with product engine capabilities in your business. We believe that continuous improvement of your processes will be treated like commercial products with end users, metrics, iterative design. We believe the tools will be built, owned, and configured to your specific workflow — not licensed in perpetuity from vendors with their own roadmaps. And we believe that your people are going to be totally different too — AI-fluent, product-minded, hired and trained to build, not just one-time training programs around the next AI tool that then fades away. The key point on this slide is that we don't believe any of these things can just be bought.

Hey Bradford, real quick — so Imran, who's on our actual technology/coding side, is tied up and can't join. He asked if we might be able to record the meeting. Would you guys have a problem with that?

Okay.

It might not let me because Allison, you scheduled the meeting — and I think you have to be a director to be able to record. [Trying to figure out recording permissions.] It's not letting me. Allison, you might be able to do the transcription and then it'll at least have that output.

We do meeting notes and transcriptions of all our meetings as well. So we can send meeting notes and a transcript.

Yeah, you'd have to do it from your end, Bradford, if you don't mind.

Yeah, meeting notes, transcription — we'll send the deck, and there is a demo that he can watch as well.

Okay, perfect. Let's do that.

So — none of these things can be bought. They have to be built by you, inside your company. Because we don't believe you can outsource transformation. And that's why we believe this moment requires a different kind of partner. If you think about all your legacy partners — agencies, consultancies, even SaaS vendors where you outsource a lot of the technology — they're set up to do the work for you and to hold on to that work. And candidly, not share any of the capability that they've developed with you back. We believe that this moment requires a different type of partner, one that is built to work with you and then step back. Because again, you can't outsource becoming a new type of organization. And so the way we've thought of this company is really designed from day one for your independence.

This is why we built Genflare. We believe it doesn't change what your commercial team does. It changes what it becomes. We believe what matters most isn't the tools you deploy but the capability you build around them. And we believe that our role should evolve as you do. First, we catalyze, build, and accelerate your work and journey. Then we scout the frontier ahead. And maybe just a little bit of explanation — folks will often ask, "Okay, really? Are you going to share all your processes and methods and the code to us? Because if that's really your model, will you have a job or a company in one year or three or five years?" And my response is: we believe so. Because what's happening right now in the market is akin to the Big Bang theory. The universe of AI opportunity is expanding faster than any person or company could possibly explore. So there will always be a need for folks like us — advisors who are on the front edge of the frontier, bringing back the goods to the mothership once your foundation has been set up.

You know who we are — we've given you our backgrounds. Maybe the key point is that we have very deep domain expertise in specialty pharma, especially technology and especially in consumer engagement. But we're also operators who have built and builders who have operated. The intersection of those three things, we think, is a rarefied group of folks who partner with pharma today. We bring a vetted network of folks depending on the project or engagement — specialty pharma operators in specific domains. We bring a capability to do full-stack AI code generation. And the third piece is this pendulum of build versus buy. Everyone talks about the SaaS apocalypse where all the enterprise software is going to go away because enterprises are building their own custom applications. We think there's a lot of truth to that. Today your mix is probably 70 to 80% buy, 20 to 30% build. We think that flips. But there will always be a need for specialized domain-specific AI vendors, and we've vetted those companies and know how to recommend depending on the use case.

This is a bit of a case study. We started with a mid-size growth-stage life sciences company, simply by rescuing a failed chatbot. They had a chatbot going for six months, it wasn't doing anything. We came in to redesign it as a digital agent to solve a commercial challenge. We did that in three months. We saw a 2x connection rate in half the time in terms of qualifying and connecting prospective patients to human care navigators, and a 3x downstream workflow conversion in the diagnostic pathway. The CEO saw that and said, "I want every single touchpoint along our patient journey to be done with a digital agent before touching a human." So we've now created that entire roadmap — I think we're on use case five or six in six months. That trickled into enterprise AI activation: company-wide rollout of tools, a commercial team keynote, AI-enabled workflows for the sales team, and a migration to Claude from ChatGPT. Then this last phase is where we accelerate the build strategy — standing up internal capability to build enterprise solutions where it doesn't make sense to buy.

There are three ways we engage. Advisory is the higher-level strategic area — helping inspire and orient executives around AI, using it as a strategic thought partner, which trickles down into execution. All of this is tilted towards execution, not strategy for strategy's sake. The second area is Signals — looking at publicly available information that pharma commercial teams don't have time to look at or aren't able to derive insights from. We have an AI pipeline that surfaces that information in a dashboard. Within patient access, this can be formulary and state Medicaid laws, hiring practices in your competition. The third area is Build — how do we stand up the strategy, process, and capabilities to accelerate enterprise custom application build for specialty pharma workflows.

I'm going to skip into a double-click on Build. AI has rewritten the economics of custom software. Small teams now ship in weeks what used to take large ones a year. What used to cost millions can now cost tens or hundreds of thousands. And what used to live in vendor systems should now live in your stack. The opportunity to build has expanded dramatically. But that comes with a caveat — just because you can build a bunch of stuff doesn't mean you know what to build or where to point those resources. That part has gotten harder.

We believe building properly requires more than incredible engineering. It requires the strategy — driving executive alignment, understanding the business case, the ROI, knowing how to scope a build. The second piece is product — taking what your case managers are doing and translating that into detailed plans. User stories, edge cases, requirements, PRDs. And the third piece is engineering. Having Silicon Valley pedigree, full-stack AI engineers is a different thing. My assessment is that most pharma companies don't have any of them. And the win is having all three together. I would say further that it's that middle product piece that many do not have. There's no enterprise product management function taking strategy and breaking it down into detailed chunks of work that an AI coding capability can start building.

Here's a case study — from strategic question to PA intake automation. The scope was large: their entire prior authorization process from intake to packet creation to appeals to competitive intelligence around payers. "It's all a mess. What should we do?" So there was a strategy component — let's understand all of it and align around a chunk of work that makes the most sense to tackle. The product piece — we mapped friction in the workflow and translated that into a working PRD, which became the basis for the RFP. We believe that level of specificity helps you find the right partner. In the RFP process we had SaaS vendors as well as an AI-native code generation company that transfers capability to the client. And what ended up happening is before the questions were even done by the other vendors, the AI-native company had created working software and showed us the demo. This process took one week of strategy alignment, one week of product planning, and two weeks of building to having production-grade working software. And I should add — we ourselves, Genflare, are AI-native. It's not just the coding that's juiced with AI, it's the strategy and the product piece. That's the only way we could have gotten to this level of acceleration across all three phases.

I have artifacts here that I can just flip through and I'll send them in follow-up. We've blinded these deliverables using our fictitious drug Lumera, a severe asthma specialty biologic. This is the strategy output document — we looked at nine different systems and eight different workflows, two distinct automation domains. One that's rule-based and high-volume intake, and one that's less repeatable because it requires more judgment. We should start with the one that's highly rule-based. Intake: not the biggest opportunity, but the right proving ground — rule-based, high-volume, low risk, immediately measurable, and it unlocks downstream work. We broke intake into five phases based on looking at the workflow: the document pipeline and fax — documents coming in, how to make sense of them, split them, sort them, route them into SharePoint. Extracting key information into the CRM. Routing and specialist handoff — once all the key pieces are in, the specialist is notified and the packet is assembled. And exception handling.

[Jessica] And Bradford, you guys did a lot of this breakdown work — user story generation, step by step — it was all done through AI also, correct?

Yes.

[Patrick] Like the intake summarization, breakdown into steps and user stories, etc. Which is again another accelerator. And just — John, I think part of what they're describing, this build-versus-buy component, is like the whole discussion we had all along when we were building our own CRM. We never wanted to be beholden to a vendor and have them own our IP and have to try to extract stuff. So we did it ourselves — it was expensive and it took time, and we kind of unwound that. Now we're kind of going back to the old way. I feel like this has been a yin and yang back and forth for Amgen's overarching CRM strategy for a while.

Yeah, totally. That's part of the benefit of our work. We are AI-native ourselves. If you just looked at us as consultants helping you figure out whatever, we believe we're 10x more productive than any other consultant that you might bring in who is not AI-native. We have all the pipeline and tricks of the trade. Anyway, I won't go through all this, but you can see it all and we'll share in follow-up.

And then lastly — we do have a Loom video of one of the demos. There were seven demo videos created; we didn't rebrand all of them, but the intake one we can share after the fact. These are just screenshots. You can see the PA specialist dashboard view — "for my work for the day, what do I need to do." The AI pipeline of things happening on behalf of the specialist: download attachments, split and classify pages, extract patient data, match or create a new patient. Triggers for manual flags — like when a date of birth is missing from documentation. So the human workflow is really exception-based. Once the specialist can move on, this is a summary of all the PA packet criteria we're waiting on — documents come in over days or weeks, and once we have everything, you approve it and it routes to the next specialist. And then there's a pretty cool analytics component — we track the efficiency and correctness of the AI pipeline itself. 80% passing, 20% not — over time you tune and improve it.

Anyway — it was impressive. By the Q&A session, it was working software. Not integrated into the client stack yet because they didn't have access, but it was pretty eye-opening how fast things can move once you have clarity on what you're trying to build. The last slide is just how to start. Patrick had shared that there are a lot of cooks in the kitchen on this work at Amgen. Politically, is it even feasible to cleave off a chunk of work and try an AI-native build? But I think there are other ways to achieve our aims — opening people's eyes around how fast things can move, while also holding people accountable. Could we build a POC on a chunk of work that's already in flight or has taken three months to just get to product requirements? "Hey guys, we have working software as a prototype, done in four weeks. Why don't you take it and use it as your wireframes to accelerate your work?" And Greg and I are full-stack executives — it's not just the build and technological component, it's the people part. Driving alignment, visioning, the narrative, being sneaky but respectful in terms of getting results at an executive level.

[Patrick] That tagline — that's a great tagline, "sneaky but respectful." Just for background, I had an opportunity to connect with Bradford and Greg — both our conversation at Asembia and subsequently — I've been clear that what they bring to the table is really important and I believe could really accelerate what we're trying to do. But we've got a number of vendors already involved, and the political landscape may be very difficult if we decided to engage them as an assist here. Depending on where this goes, we'd have to be really thoughtful about how to engage and how to bring them in with our current partners. People are going to feel like their toes are getting stepped on. They are aware of that as a challenge for us.

[John] I think it's a very good flag. I can think of three or four things off the top of my head, but I agree — that is a potential rate limiter. To the extent that one of those teams or vendors is building today or moving the needle but maybe not fast enough, or the thesis is not as much AI in the mix — there is a move here, we should talk about it. But to what extent can we be open and friendly and on the same side as that team?

[Patrick] I think part of the problem is the traditional vendor approach doesn't necessarily benefit from acceleration through AI. We're paying them based on hours of consultation, and the two models will be in conflict with one another.

I was thinking more — just to be clear about the breadth of teams you're talking about — but if one of them is an internal team, for example, or an IT team that's starting to use AI, that's a natural place for us to lean in. We could augment their team, bring in extra capabilities, surround them with good product design pedigree that you would do in commercial software launches. But I agree completely — it's completely oil and water to go help a SaaS vendor do it better. They would have to jump the curve and change their whole business and let all their venture capitalists know they're changing horses.

[Jessica] I'm processing a couple ideas that have come to mind. Patrick, John, Allison — it may be worth us brainstorming a little bit on what could we potentially do with you guys as a proof of concept. It may not necessarily be the most impactful or life-changing, but it gives us an opportunity to test it out with expansion to other use cases. I'm also wondering — does this help with some of our current limitations in the short term? Where we can look at some of those manual processes and — it's not going to live on for years, but if we can get it out the door quickly, that takes away the challenge of continuing to invest time and money on a platform that we're going to decommission but is incredibly painful and inefficient.

[John] Certainly painful. To be having this growth and lots of technology out there, and we're stuck on something that's antiquated. I was literally bumping heads to get to that same spot, Jess — if we're looking for something to demonstrate the value here in an effort to help sway the political challenges for broader adoption, in my mind the exact right spot to point this at is that interim state we're in. Because there are clearly enough challenges there where we can highlight a few for consideration. So totally aligned on us circling up to brainstorm and then circling back with you guys to get your thoughts on what we put together.

And you spoke to this earlier — I'm just trying to get a sense of how quickly could you launch based on the use cases? Some averages?

I think four to six weeks is a good target. Of course, the first thing you do is look for a chunk of it you can deliver that has value, that stands up. If what we're trying to do is verify a path or an optionality, you want it to be meaty enough that it really pulls off the difficult work in a clear way without doing all of it. But I think we should really look to keep it in four to six weeks — and that's from product requirements documentation, dev, then test and deploy. I would just add that it's the integration component that's going to slow everything down. So I don't know what system you're on, or if there are some easy ways to not integrate — there are definitely parts of the workflow where you don't need to be integrated, or you could do a CSV flat-file drop and that's how we run for a little bit just to show that the workflow component works and drives value.

[Jessica] Does it make sense then for us on the Amgen side to circle back — ideally if we can find some time this week — and brainstorm on some of the ideas that we think could be possible, and then come back to you guys and give a little bit more context? Like, "These are the five potential problems that we have — how could we potentially address something like this? What would you advise?" And then we can see if any of those seem like good POC situations to start to tackle. Does that give us the next tangible step here?

[John] I like that a lot.

I would just say that to the extent that you record that session so you've got lots of notes around it, we could even get a jump start. If we see the context of your discussions and you've gotten down to four or five and we know that ahead of the meeting, we can come to the meeting with some structure and a way of thinking about those — so that we come out of that meeting with some concrete results. Like, "These are the top two we should think about." But I love that. I think that's the right next step.

Housekeeping items — we do need an NDA in place. I don't believe we have one, Patrick.

[Patrick] We do have a boilerplate NDA that we could probably use from our legal team if that's helpful. Probably faster to get it through on our side if we use our template.

We'll take that too.

[John] Really appreciate it, guys. This is exciting. Not for anything, considering the volume of change that we as an organization are being asked to make over the next two years, this actually has me breathing a little bit easier. And I think I've told you guys this in previous conversations — part of the way that we're scored on performance and part of our corporate ethos and direction is finding better ways to use AI to do what we do faster. This fits right in with our corporate direction right now. I don't know if you've seen the press release — we actually bought a bunch of servers and did a partnership with NVIDIA for drug discovery and some other things. Amgen's kind of all-in on what we can do more of and more quickly with AI. Which is good.

It is good. That's another stanchion in this sprint — another optical strategic connection to the mothership. Well, awesome. We really appreciate the opportunity to learn and grow and experiment with you. We'll look forward to the next session. Good to meet all of you.

We ended five minutes before the end of the meeting.

Wild.

Look — AI is already making us more efficient.

[Patrick] One thing — I think you guys can see there's a lot of interest here in proving out a use case. Before we get too far ahead of ourselves, we'll get the NDA in place, but we should talk about how you guys price too, so we understand what we're looking at from a cost standpoint. Is it billed hourly, or project-based? Because there will — we will have to go make those asks from a budget standpoint. We don't just have extra budget sitting around for new test cases. It will be an incremental buy-up on our side. So understanding what hurdles are coming down the pike as we move further along.

We can address that in the next meeting — at least just the model.

Awesome. Thanks, you both. Appreciate it.