USA
UK
Canada
France
Germany
Italy
Spain
On the latest episode of the Data Humanized Podcast, Host Mark Palmer interviewed Fernando Schwartz. Fernando is the Vice President (VP) of Data Science & ML Engineering at ADP. In the episode, the pair discuss how to:
- Use generative AI and large language models (LLMs) at the enterprise level
- Integrate your people, process, and technology organization-wide
Keep reading for the top takeaways from Mark and Fernando’s conversation.
Takeaway #1: Make content AI-ready for maximum impact
Generative AI, while incredibly efficient, doesn’t always grasp the full context or nuances in data and relationships the way humans do.
That’s why it’s essential to prep the content you feed into generative AI and LLMs.
Fernando shared that generative AI and LLM output is only as good as the inputs you give the technology.
To generate high-quality output, he recommends using ontologies or tags to add context.
Structuring knowledge for ontologies has traditionally been human-intensive and slow, but Fernando and his team have generative AI take on the heavy lifting.
We want to be able to iterate quickly and scale fast. So traditionally when you build ontologies [it] is very intensive, human work that's required to build these things. [Instead,] we use generative AI to do that and then use humans [to check the output].
Takeaway #2: Layer the use of generative AI and LLM applications
Fernando also recommends making content inputs AI-ready by doubling down on generative AI and LLM use. He shared that teams at ADP use LLMs to not only reformat data but also to evaluate if reformatted outputs are better than original inputs.
We're using LLMs to reformat content, but we’re also using LLMs to determine whether this new reformatted content is better than the previous formatted content. This is using what's called LLM as a judge. When you start doing that then you can iterate quickly because you can sort of automate everything.
Takeaway #3: Generative AI needs a human touch
There’s no doubt that generative AI has transformed business operations with the speed, scalability, and efficiency it offers. However, it still requires human intervention to reach its full potential.
During the podcast episode, Fernando stressed the critical role people have to play in enterprise generative AI applications.
He noted that employees are needed to increase the accuracy of generative AI outputs — especially at organizations like ADP where data accuracy is essential.
You think of the people using [the technology] as aliens. They actually understand the context, the data, and the relationships between and among them.
Takeaway #4: Equip employees with AI and LLM skills
AI and LLM technologies are expected to increase exponentially in enterprises like ADP. To keep up with the evolution of these technologies, technical and non-technical employees alike need proper upskilling.
Fernando shared his thoughts on the need for AI and LLM workforce skills on the podcast:
Before people can build useful things they need to be acquainted with [the technologies]. Everyone needs to understand how to use [generative AI] and even LLMs [as they] get more accurate.
Takeaway #5: Apply generative AI to your customer experience
As enterprises like ADP explore how to integrate generative AI, customer experience workflows are ready for its application.
This use case is not just about efficiency, but about rethinking the entire customer journey and leveraging generative AI to understand and anticipate customer needs.
For example, ADP has leveraged AI to better serve its customers when they use the search bar. Their application of AI picks up on potential customer needs and proactively suggests a solution.
During the podcast, Fernando reflected on the potential of generative AI to improve customer experiences:
Generative AI is a new channel to reach our clients. We're trying to get a really quick understanding of [customer] intent so when they go into the product, we quickly realize what they want to do and then help them get there. So we use generative AI to remove friction.
Takeaway #6: Communicate the bigger picture to your data teams
A challenge enterprises face is integrating data from multiple sources.
When teams collect and analyze data, it’s typically done with a specific use case or department goal in mind.
However, this narrow perspective overlooks the value data can bring when it's integrated into a broader organizational context.
Fernando highlighted the importance of telling data teams how their work fits into the larger strategy on the podcast:
That mentality of thinking okay my data is mine but it's also going to contribute to a bigger pile of data that can add value to the organization. This is something that takes a little bit of understanding.
Fernando said data teams must understand that their contributions are not just department-specific but part of a larger repository of data that fuels insights organization-wide.
Bringing it all together: The future of AI and LLMs in enterprises
Generative AI and LLMs are a game-changer for enterprise operations. These technologies improve organizational outcomes by:
- Automating tasks that traditionally required a lot of manual effort
- Empowering employees to focus on high-value tasks
As these technologies become more embedded in organizations, people leaders must focus on thoughtful application, workflow integration, and developing the digital skills of employees.
Partner with Correlation One to equip your workforce with advanced AI and LLM skills. Get started today.