Blog

Why AI isn't crossing commercial functions; 3 questions for Neurelis' Brett Ramos

Written by Marc Iskowitz | Nov 19 2026

Welcome to edition fifteen of The Beghou Biopharma Review by Marc Iskowitz.

Two findings capture the gap between AI ambition and operational impact.

  1. In late 2024, 85% of specialty market commercial leaders expected their organizations to implement GenAI across multiple functions within the following year.1
  2. In our 2026 Commercialization That Works research, just 10% of a broader commercial sample reported measurable AI impact across multiple workflows.2

The studies are not directly comparable, but together they point to a clear reality: Expectations have moved considerably faster than execution.

In other words, in late 2024 when organizations still had limited experience with generative or agentic AI, few realized what it would take to scale across multiple commercialization workflows, and according to this year’s Commercialization That Works findings, they're still trying to figure it out. I noted that teams are being "asked to bolt AI onto foundations that aren’t coherent, composable, or cohesive." Let's focus on the third of those 3 Cs and explain why a lack of cohesion is making enterprise AI an uphill battle, both in terms of tools and the way the technology is being rolled out.

Ensuring AI fits within workflows: the tech side

When I say "cohesion," I refer to the extent to which users, workflows, data, and technology can work together in practice. According to our data, tools and processes are not solving people’s day-to-day challenges:3 

  • Only 7% strongly agreed they can translate decisions into execution across core workflows without heavy manual rework. 

  • Only 7% strongly agreed their workflows are streamlined and standardized enough to avoid manual handoffs. 

  • Compared with 24 months ago, only 5% said it’s much easier to keep work from depending on manual handoffs or specific people who know how things get done. 

The takeaway: Layering AI onto an out-of-sync commercial foundation could be one reason companies are struggling with its scaling and operationalization.

What to do now: In order to facilitate cohesion in their commercial model, organizations should do the following:

  1. Design the technology and the operating model together, rather than having the tech team build something and asking users to validate it at the end.   

  2. Design around how end users work, involving them early and incorporating their feedback as the solution is used.
  3. Prioritize the AI opportunities that matter most, using a decision matrix to weigh how often workflow decisions occur and how critical they are.

Change management: the people side

Pharma’s uneven adoption begs the question: Have we created the conditions for people to use AI confidently in their work? Beghou Partner Nicole Ventrone wrote two pieces on turning AI interest into durable adoption in life sciences commercialization, drawing on a client project and research with 100 biopharma professionals that she co-presented at PMSA.4

In Pharmaceutical Executive, Nicole notes how compliance uncertainty turned out to be the leading barrier to adoption, followed by data quality, training, and a lack of approved use cases. As she points out, the workflow deserves as much attention as the technology.

In another piece for The AI Journal, Nicole shares the example of an AI project that stayed sidelined because no time was cleared for learning or workflow redesign. The case highlights the distinction between enthusiasm for AI and the operating capacity for meaningful implementation.

Indeed, adoption can improve when teams explicitly protect learning time, assign a workflow owner, and measure whether the new process streamlines existing work rather than simply adding another demand.

👉Check out Nicole’s AI Enablement report for additional findings.

3 questions for... Neurelis' Brett Ramos

In a recent episode of The Space Life Sciences Podcast, I sat down with Brett Ramos, who co-led the above AI adoption study with Nicole earlier this year. Brett, who subsequently joined Neurelis as Senior Director of Commercial Operations and Analytics, discussed why organizations need to better understand employees' behavioral barriers, demonstrate value, and make AI easier to incorporate into everyday work.

The following questions and answers have been edited for length.

Q: We routinely segment external customers and personalize messaging to maximize engagement. Why haven't companies applied the same approach to AI adoption internally?

Brett: We've long been familiar with segmenting our customers, but we haven't focused as much on understanding the barriers within our own organizations. People aren't always willing to look in the mirror as closely as they examine their customers. And those barriers vary. Someone might feel overwhelmed by having to learn new technology, while someone else might be concerned about compliance or return on investment. Understanding those different mindsets creates an opportunity to tailor how we encourage adoption.

Q. Can you share an example of how you overcame organizational barriers to AI implementation?

Brett:  We developed AI-powered patient-finding alerts designed to help identify patients who might otherwise be missed in traditional claims data. Initially, we intentionally spent considerable time demonstrating to the users that the algorithms worked, sharing videos and examples of successful results. After three or four quarters of delivering the alerts, that initial work paid off. Instead of asking people to use them, we started hearing, "When's the next AI alert coming out?" Adoption had shifted from a push to a pull. Showing consistent wins helped people overcome their initial skepticism.

Q. If AI adoption is largely a change management challenge, what still needs to improve on the technology side?

Brett:  The technology needs to be built into people's existing processes. Consider tools like Claude Code, which many people have traditionally accessed through a command-line interface. For employees unfamiliar with that environment, the response may simply be, "No, thank you." But more graphical interfaces and low-code or no-code platforms are making sophisticated AI capabilities accessible to people without programming experience. Making these tools easier to use—and less intimidating—is an important part of encouraging adoption.

👂 Listen to the podcast episode.

What's next...

At JPM Week in San Francisco (January 11-14), Beghou Partners Jenny Herritz, Nicole Ventrone, Robert Rouse, and Harsh Kulkarni will be hosting a “Boba with Beghou” meetup. Contact me at the email below to RSVP.

That’s it for now. If you’re looking for more context on any of these stories, drop me a line at marc.iskowitz@beghou.com

Thanks, as always, for reading,

Marc Iskowitz

 

1 Refreshed and refocused for 2025 survey (December 2024); n=55 U.S.-based commercial operations, analytics, technology, and marketing professionals working in specialty markets

2 Commercialization That Works 2026 survey (August/September 2026); n=120 U.S.-based commercial operations and data and analytics/commercial IT professionals

3 Commercialization That Works 2026 survey (August/September 2026); n=180 U.S.-based commercial operations, data and analytics/commercial IT, HCP and patient marketing, and market access professionals

4 Commercialization That Works: AI Enablement survey (April 2026); n=100 U.S.-based, pharma and biopharma professionals