How to Scale AI Learning with PayPal: AI Foundations

コンテンツ概要

Many AI learning programs start strong but lose momentum when teams lack clear direction on what skills actually matter. Drawing on insights from Udemy’s webinar with PayPal, this article explores how their team built a scalable AI learning strategy by focusing first on strong foundations, helping employees understand ethical use, core AI concepts, and how to apply tools effectively in their daily work.

There’s a moment many L&D leaders are familiar with right now. Leadership asks how the organization is progressing on AI readiness. You have a program in place, maybe some curated content, a few learning paths, a platform people can access. And yet, something isn’t quite adding up. Engagement is hard to sustain, the business wants more, and you’re not entirely sure where to go next.

If that sounds familiar, you’re not alone, and you’re not behind.

Recently, I sat down with Brittany Ward, Director of Enterprise Learning & Development at PayPal, for an honest conversation about how PayPal is building and scaling AI enablement across their organization. I appreciate that Brittany didn’t show only a polished success story. She showed real experience, the pivots, the lessons, and the framework that has started to shape the way PayPal thinks about AI learning on Udemy.

What emerged from that conversation was a three-pillar framework: AI foundations, AI fluency, and becoming AI native. This post focuses on just one of those pillars, the one that Brittany’s team identified as the essential starting point before anything else can work: AI foundations.

Why AI learning programs lose momentum before they scale

Many AI learning programs fail because of unclear focus. A pattern that many L&D teams will recognize after launching a new AI initiative. People are curious, engagement is high, and there’s real energy around AI upskilling and learning. Then, a few months in, that energy quietly fades and new initiatives compete for attention. The program is still there, but fewer people are actively engaging with it.

When organizations move quickly to get something out there, pulling together curated content, building broad learning paths, giving people access to everything they might ever need, the result can feel more overwhelming than helpful. There’s no clear signal about where to start or what actually matters most. Without that, even well-intentioned programs lose traction.

PayPal experienced this firsthand. In the full webinar, Brittany is refreshingly candid about what happened, and the specific pivot her team made to change course.

A quick look at PayPal’s three-pillar framework

Before we get into the details, it helps to understand the shape of what PayPal has built. After an initial launch that generated strong early engagement, Brittany’s team stepped back and asked a harder question: “Are we actually building capability, or are we just generating activity?”

That question led them to anchor their AI enablement approach in three distinct pillars.

Pillar one: AI foundations:  This pillar is about building the knowledge base that people need before they ever open a tool. It covers the concepts, the context, and the guardrails that help individuals engage with AI confidently and responsibly.

Pillar two: AI fluency: This is where knowledge becomes application. AI fluency is about getting people hands-on inside the tools their organization has already committed to, and making sure the learning doesn’t feel separate from the work they’re doing every day.

Pillar three: AI native: This pillar focuses on leadership. Becoming AI native means enabling leaders to model, govern, and champion AI adoption across teams, while thinking seriously about what work redesign looks like when AI is part of the picture.

Today’s blog focuses on the first pillar of the framework. It’s where Brittany’s team starts every conversation about AI enablement, and it’s where we’d encourage you to start too.

Pillar one: AI foundations 

AI foundations are about giving people enough shared understanding so they can engage with the tools, ask better questions, and apply what they learn in a way that actually moves work forward.

When people don’t have that foundation, a few things tend to happen: they approach tools with uncertainty, produce inconsistent outputs, and are often unsure whether what they’re doing is appropriate or effective. Over time, skepticism builds, not because AI isn’t useful, but because no one helped them understand how to use it well.

PayPal’s foundations pillar is built around three core areas that help fortify this key part of the framework.

Ethical AI use

Before anyone at PayPal is expected to use AI in their work, the expectation is that they understand what responsible use of AI looks like. This is especially important in a financial services organization, where the stakes around data, compliance, and customer trust are high.

Ethical use is a lens that shapes how people approach every interaction with an AI tool, what information they share, how they evaluate outputs, and when they apply human judgment rather than defaulting to what the tool produces.

Understanding LLMs and core AI concepts

There’s a lot of confusion in most organizations about what AI actually is, and what it isn’t. Large language models are often either mystified or oversimplified and neither helps.

PayPal’s foundations work includes helping people understand, at a practical level, what an LLM is and how it works. Not at an engineering depth, but enough that people can engage with AI tools thoughtfully rather than treating them like a magic box that either impresses or disappoints.

When people understand the basics of how these models work, how they’re trained, what they’re good at, and where they fall short, they use them more effectively. They’re also better at spotting when an output doesn’t look right, which matters a great deal in any organization where accuracy is important.

Tools, techniques, and knowing the difference

Not every AI tool does the same thing. The technique that works well in one tool may not translate to another. This is something L&D teams often underestimate when building AI learning programs.

Part of building a solid foundation is helping people understand the landscape of tools their organization uses, what each one is best suited for, and which AI techniques, including prompting approaches, help them get better results. This is a form of orientation rather than deep tool training. It helps people know where to go and what to expect before they’re sitting in a tool with a real task in front of them.

Why AI foundations have to come first

It might seem obvious that you’d build foundations before fluency. But in practice, many organizations skip this step, or rush through it, because there’s pressure to get people doing things, not just learning things.

That pressure is understandable. But skipping foundations creates problems that show up later, and they’re harder to fix once people have already formed habits around tools they don’t fully understand.

The content-first trap

One of the most common patterns in AI learning programs is what might be called the content-first approach. Teams build or curate a library of content, create a place people can access it, and make it available. The thinking is simple: give people the resources, and they’ll learn.

The challenge is that content without context doesn’t build capability. It builds access.

Brittany put it directly in the webinar: the content-first model doesn’t scale, and AI is making that clearer than ever. When the pace of change means some content is outdated within two to three months of being published, the idea that you can curate your way to AI readiness starts to break down quickly.

What changes when foundations come first

When people have a shared foundation, everything that follows works better.

They engage with tools differently because they understand what they’re working with. They apply AI techniques with more confidence because they know what good output looks like. They raise appropriate concerns because they understand what responsible use requires.

Perhaps most importantly, learning stops feeling like an event that happens before the real work begins. It becomes part of how people think about the work itself.

Brittany shares how PayPal is embedding this kind of learning directly into the tools their teams use every day, and what they’re finding out about the difference between learning that sticks and learning that doesn’t.

Learn more about how PayPal scales AI learning

In the full Udemy x PayPal webinar, Brittany Ward shares what comes after foundations, including how PayPal is moving from knowledge to hands-on application inside the tools their teams use every day, how they’re using backend telemetry from vendors to understand whether people are actually applying what they’ve learned, and what happens when the data says the learning isn’t landing.

You’ll also hear about the work happening at the leadership level, including how Brittany is personally using AI agents to manage her own work, what that has taught her about the future of team leadership, and how PayPal is thinking about building AI capability from both ends of the organization at once.

You’ll hear the honest version of the journey, the moments where something wasn’t working, the decision to pivot, and what that process actually looks like inside a large, fast-moving organization.

Watch the full webinar and learn the rest of the pillars and how PayPal scales AI learning.