In the latest episode of Deployed, we talked with Eliza Cabrera (Principal Product Manager) and Beau Lyddon (Principal Engineer) who are building Workday’s AI products to discuss their journey building and scaling AI products for enterprise customers, and in an enterprise environment.
As one of the leading HR and finance software providers serving 60% of the Fortune 500, Workday has valuable insights to share about making AI work at the enterprise level. They’re also a Fortune 500 company themselves, and lots of the discussion focuses on what they’ve done to build a platform that helps multiple product development teams create AI features for customers.
Big themes center around:
Figuring out what to build for enterprise customers: The combination not just of what’s possible and how to deliver value, but also what’s realistic given customers’ concerns about data privacy and compliance, level of familiarity with AI, etc.
Breaking ground and getting AI products to market in the enterprise: The mix of product, business, and platform considerations that went into getting the first products live, and then being able to take on more ambitious generative AI products – like the recently-announced Workday Illuminate platform and AI Assistant.
Building capacity and scaling adoption of generative AI across Workday products: Discussion on a range of topics that matter for building in large product and engineering organizations, from making it easier for people to do prompt engineering to building services that lock down data access appropriately and make it possible to build secure enterprise products.
Check out the full episode here on YouTube, Spotify, or Apple Podcasts. Some of our favorite highlights are below.
Scaling best practices across product & engineering teams
Eliza & Beau’s story is interesting, and common at big companies. Eliza’s team was responsible for building the first generative AI features at Workday, and then Beau’s platform team partnered with them to figure out what abstractions were needed to help other product teams build more AI features. Some of that turned into code and tools provided by Beau’s team, and Eliza led the creation of guidelines and a playbook for other product teams to follow.
In this clip Beau talks about some of their learnings and how they discovered which elements were needed at the platform level.
On Creating Easy-to-Use AI Features
Another learning that they share — and that we’ve seen repeated across dozens of companies – is that the most useful AI features are embedded in existing product features. Separate chatbot interfaces have their place, but they also require customers to think about what to say, learn how to prompt LLMs effectively, etc.
Eliza and Beau talk in this clip about their experience building prompts into various parts of the product – like “Translate” or “Summarize” buttons — and why they’ve been helpful for customers.
Making AI Agents Work In The Enterprise
The realities of building AI features at the enterprise level are much different than they are for small startups or demos on Twitter. Beau shares some of the details about where Workday’s platform is investing – for instance, building services that protect access to data so that agents don’t leak information that they shouldn’t. For instance, a manager should be able to ask questions about their team’s salaries, but employees on the team shouldn’t be able to learn about each other’s salaries,
These are the kinds of practical realities that so many enterprise software teams are trying to solve for as they invest in generative AI.
Parting Recommendation: Get Your Hands Dirty
We asked Eliza & Beau for their parting recommendations to other teams building in environments like Workday. Eliza’s resonates: Use AI yourself as much as you can. We’ve found this ourselves and often share the same advice. Generative AI is like a new building material, and as builders try to figure out how to use it in their products, there’s perhaps nothing more valuable that getting a native sense for how it behaves.
Looking Ahead
Both Eliza and Beau emphasized that while the technology is exciting, the hard work is in making it truly valuable for enterprise customers. As Beau noted, "The playing around is fun...but figuring out how to get from fun to actually delivering customer value is the place everybody is in right now." That's where Freeplay comes in. If you're building enterprise AI products, we'd love to talk about how we're helping AI product teams build great customer experience.
And for organizations looking to learn more about Workday's approach to AI, visit workday.com/AI or check out their AI Masterclass content. Their new Assistant product, powered by their Illuminate AI platform, is available to early adopters now.
Want to hear more conversations like this? Subscribe to Deployed on Spotify, Apple Podcasts or YouTube.
