At Version 1’s Women in Tech Leadership: Pharmaceuticals, MedTech & Life Sciences event in Princeton, New Jersey, Kristi Lecher, Executive Director of Digital Platforms, Data and AI at PCI Pharma Services, shared hard-won lessons on scaling digital transformation and AI in a regulated industry. From building a customer-facing transparency platform to navigating the realities of enterprise AI adoption, Kristi shares her biggest learnings along the way.

Kristi Lecher didn’t start her career in technology. Fifteen years ago, she joined PCI Pharma Services as a project manager, running procurement teams, supply chain operations, and project management on the operational front line of the business. Today, she leads digital platforms, data strategy, and AI for the global Contract Development and Manufacturing Organisation (CDMO) that has grown significantly over the past 15 years.

Her path from operations to technology leadership was the starting point for a candid fireside conversation with Nick McFadden, Version 1’s Regional Director for North America, at a recent pharma and life sciences leadership event. What followed was a practical, honest look at the challenges technology leaders face in regulated industries as they try to move from AI ambition to enterprise reality.

Two speakers seated at a small table address an audience at Version1's Women in Tech Leadership NJ: Pharmaceuticals, MedTech & Life Sciences event, with attendees listening in an indoor setting.

When operational knowledge becomes a superpower

When Lecher was asked to take over PCI’s customer-facing digital platform, pci | bridgeTM, she describes the moment with characteristic honesty: “I was really scared because I thought – I’ve never done this before. I’m not a tech person.”

But the leap from project management to product management turned out to be less of a gap than she expected. Years spent sitting across the table from clients, managing supply chains, and understanding what operational teams actually needed day-to-day gave her something a purely technical background might not have: an intuitive understanding of what users were trying to solve.

“What I was trying to do was solve the solutions of my user base,” she explains. “To be able to translate the operational knowledge into the solution and driving feature sets and capabilities, it became a lot more native to me than I really ever expected.”

That operational lens now shapes how she approaches everything from data strategy to AI adoption. In an industry where the distance between a technology investment and its impact on a production floor, a patient, or a regulatory submission can feel enormous, that perspective matters.

The platform that changed the conversation

pci | bridgeTM, the customer-facing platform Lecher oversees, offers pharmaceutical clients real-time visibility into their portfolio of work at PCI. That includes open order management, production status, shipment tracking, inventory, and document collaboration. It represents a fundamental shift in how a CDMO relates to its customers.

“It was kind of a shift for us to go from a black box to a glass house,” Lecher says.

In practical terms, PCI’s pharmaceutical clients historically had limited visibility into what was happening with their outsourced work. They would hand over their products and wait for updates. pci | bridgeTM changed that by giving clients 24/7, real-time access to the status of their operations. The “glass house” means PCI is now willingly showing clients everything, even when the data isn’t flattering.

“Not a lot of organisations are willing to be that open. You’re sometimes going to see our warts, and that’s just how it is.”

That willingness to show transparency has become a competitive differentiator. But it also requires careful thought about how information is presented. Every new feature addition prompts the same set of questions: are we comfortable sharing this? What’s the right context? How do we make sure clients can consume this data without it raising alarm bells out of context?

As client expectations evolve, so does the platform. Some customers now want direct integration into their own control towers and centralised data environments. Others are asking for manufacturing execution system data, including real-time line uptime, yield percentages, and scrap rates. PCI is building the data foundations to make that level of transparency possible in the future.

Where AI is actually landing in pharma

When it comes to AI adoption, Lecher sees a clear pattern across PCI’s customer base of roughly 1,400 pharmaceutical clients, ranging from early-stage biotechs to multi-billion-dollar global organisations.

The organisations struggling most, she observes, are the ones chasing the biggest, most ambitious use cases first- clinical R&D applications, drug product prediction and clinical trial optimisation. “A lot of that sort of feels like voodoo,” she says. “The organisations that are having a harder time translating it are the ones that are focused on those big picture AI use cases.”

The use cases gaining real traction are more grounded: technical writing and document summarisation, working with unstructured data across different systems, and creating workforce enablement tools that help employees spend more time on value-added activity and less on administrative tasks.

Lecher highlights three areas where she sees the strongest near-term opportunity:

  • Hands holding three people icon

    Organisational knowledge bases

    With 38 global GMP sites, a heavily regulated documentation environment, and constant workforce movement, the ability to query institutional knowledge (SOPs, work instructions, training materials, historical context) represents enormous potential. “Nobody’s cracked how you download somebody’s brain,” she says. “You have people that onboard and offboard. You have to be able to share that knowledge, but unless somebody’s been homegrown with the organisation, you can’t just plug them in day one.”

  • A circular arrow surrounding a gear icon, representing automation, continuous improvement, or process optimisation.

    Predictive maintenance and engineering enablement

    PCI’s engineering teams are exploring knowledge bases that could help mechanics on the production floor understand maintenance schedules, equipment history, and predictive maintenance signals. This directly supports the uptime of the production lines that are core to PCI’s capacity-based business model.

  • Icon

    Workflow enablement and employee productivity

    Rather than building bespoke AI solutions from scratch, Lecher sees growing momentum around adopting the native AI capabilities already embedded in existing enterprise software. These are tools employees may already be comfortable with from their personal lives, and enabling those quickly serves as a first step toward broader adoption.

Why adoption fails: the product mindset

Perhaps the most resonant thread in the conversation was Lecher’s analysis of why AI initiatives stall. She draws directly from her experience launching pci | bridgeTM.

“The intention was that we were going to build this really cool tool and everybody was going to come,” she recalls. “We heard from customers what they wanted, we built an MVP, and then trying to get over the adoption curve, the human psychology, the behaviour of adoption of tools, we were struggling. It’s not because we didn’t build a great product. It was just, how do you get people to train the behaviour to go do something different than they did before?”

That experience fundamentally shaped her approach to AI. She now treats every internal AI initiative as an internal product launch, with the same rigour around user enablement, training, and change management that she would apply to an external-facing platform.

“If you want adoption to happen, you have to take a product mindset and an enablement mindset,” she says. “If you don’t spend the time on the enablement side of things and bringing the organisation along on the journey, adoption fails. I think that’s probably why a lot of organisations get stuck in pilot purgatory. These tools are being put in and it’s being done to the organisation instead of with the organisation.”

Her team member Ahnika, a customer success manager for pci | bridgeTM, spends much of her time doing exactly this: training users, demonstrating value, and building the habits that drive sustained adoption. Lecher sees the same function as essential for any internal AI rollout.

Governance without paralysis

In a regulated industry, the tension between innovation speed and compliance rigour is constant. Lecher outlined several practical approaches PCI is taking to navigate this.

On the supplier side, PCI is building AI-specific evaluation criteria into its vendor management processes. As enterprise software providers roll out AI features (sometimes enabled by default) PCI needs the ability to assess whether those features operate in a closed loop, evaluate what controls are available, and determine whether certain capabilities can be toggled on or off. “Sometimes we’re finding that no, that’s not possible,” Lecher notes. “And other times we are.”

Client expectations add another layer. Many of PCI’s pharmaceutical customers are actively placing requirements on how PCI uses AI, particularly where it touches their proprietary data. In response, PCI has published an AI intended use statement and compliance framework that commits to keeping client data in a closed loop, maintaining human involvement, and being transparent about where AI is applied.

Lecher’s advice on working with quality, compliance, and regulatory teams is pointed: bring them in early, treat them as partners rather than police, and focus on educating them about outcomes and guardrails. “If you utilise them in that way and you educate them about what you’re trying to do, you’re trying to explain the guardrails you’re putting in place, they’re more likely to be supportive of the innovation you’re driving because they know that you’re not being irresponsible with it.”

It’s not the model. It’s everything around it.

When asked what keeps her up at night, Lecher’s answer cuts to the heart of a challenge many technology leaders will recognise.

“It’s not the model that’s the problem. It’s everything around the model. It’s how inefficient the workflow is. It’s this magical expectation that AI is going to solve a problem that the organisation hasn’t even solved yet. And if the data’s not there, you can’t necessarily manufacture that.”

This is why PCI’s data platform initiative, a project Version 1 is partnering on, is so central to Lecher’s strategy. After years of advocating for centralised master data management (she describes herself as “a choir of one” who eventually got others to join the chorus), the organisation has now funded and begun implementing a unified data platform.

Without that foundation, AI use cases will continue to be constrained by the limitations of siloed systems of record. With it, entirely new possibilities open up, from richer customer transparency through pci | bridgeTM  to more sophisticated AI applications across the enterprise.

Making the business case: connect to strategy, not technology

On the perennial challenge of securing budget for data and AI initiatives, Lecher shared a framework that has consistently worked for her.

“The biggest success I’ve had in getting budget approved is making sure that I’m staying close to the strategic projects that are happening in the organisation, the organisational goals being communicated by the C-suite,” she explains. “If I can connect the use case as an accelerator for those deliverables, or connect it to a pain point they haven’t figured out how to solve, the majority of the time I can get funding approved.”

She deliberately avoids leading with technology costs. Instead, she frames investments in terms of their connection to strategic initiatives and measurable business outcomes. And she offers a practical warning, “make sure you don’t lock yourself in on something you can’t track. We’ve definitely hit that hurdle before, thinking we’re signing up for a certain ROI, but then it really isn’t something that’s attributable.”

Build, buy, or wait?

Lecher takes a pragmatic, case-by-case approach to the build-versus-buy question. PCI has evaluated roughly ten different AI proposals, and several that initially looked promising were shelved when the ROI couldn’t be justified or when it became clear the use case would be better served after the data platform was in place.

For employee productivity use cases, she’s currently leaning toward the buy side. Leveraging off-the-shelf tools that employees are already familiar with from their personal lives, and that are increasingly becoming table stakes in the workplace. For more complex, domain-specific applications, the build path makes more sense, but only once the data foundation is there.

The key, she emphasises, is resisting the urge to jump to the most ambitious solution when a simpler enabling step might deliver faster value and build the organisational muscle for larger initiatives later.

Version 1 partners with PCI Pharma Services across enterprise applications, cloud, data, and AI. To learn more about how Version 1 supports organisations in pharma, medtech, and life sciences to navigate digital transformation and deliver real customer success, contact us or explore our life sciences capabilities.

This article is based on a fireside conversation at Version 1’s Women in Tech Leadership event, New Jersey, 2026.