AI Implementation Faces HR Data Silo Challenge

As employers continue to implement artificial intelligence technologies and use generative AI platforms that depend on company data from across the organization, many HR stakeholders doubt that these technologies will break down data silos that hold sensitive employee information and are often stored in a standalone legacy system walled off from other parts of the business.

These data silos hold a wealth of information that can unlock key insights about employee experience, performance and behavior, among other types of critical information.

However, as the cry for more data is heard, analysts like Helen Poitevin, who works in Gartner’s HR practice, have cautiously expressed their concern that trying to tap into siloed data isn’t always a useful task.

“Sometimes, data is siloed for a reason. Benefits data, for example, is siloed for a very good reason—that data is much more personal,” Poitevin said.

She added that HR leaders shouldn’t feel like they have to break down all the siloed data, nor should they think that this approach will somehow make AI easier to work with. 

“You don’t need to break down all the silos to make AI work. What HR leaders should do is think about why they implemented AI, what the purpose behind it is and which data is best to serve that purpose,” Poitevin said.  

Jimmie Lee, CEO at JLee & Associates, a Seattle-based technology firm, said one situation that has added to the complicated siloed data structure is when business unit heads across companies buy a new tool for their own department, but don’t think about how those tools work across the IT enterprise.

“You have your chief financial officer going and buying finance tools that don’t work well with the HR tools, that don’t work well with the IT tools, that don’t work well with products or services. All of these different areas of the organization are buying their own technologies, and nobody is sitting on top of them saying, ‘OK, guys, you should all technically work together,’ ” Lee said.

Lee said that every three years, there’s a new tool, a new practice and a new structure in different areas under the HR umbrella.

“Nobody has gone back and really organized the 12 different tools that the company has in their past to bring all of that data together,” he said.

There are many reasons why companies have created siloed data repositories over time. For a start, regulatory and compliance rules often prevent companies from using certain types of data. Additionally, companies haven’t paid enough attention to their growing tech stacks.

In the meantime, creating a collaborative company culture can help business unit leaders find solutions to their siloed data problem, but Lee said many company executives are skeptical about sharing their data.

“There has been a lot of distrust or cultural issues among the top leaders that they don’t want to surrender access or too much data to other executives. Company leaders want to have full control over their data,” Lee said.

According to the Zendesk Customer Experience Trends Report 2023, only 22 percent of business leaders said their teams share data well.

One company that’s looking to pull employee-related data from multiple systems is Ask-AI, headquartered in Tel Aviv, Israel.

Ask-AI is a generative AI software solution that connects to more than 50 enterprise work systems such as Salesforce, Zendesk, Confluence, Jira, Slack, Google Drive, Teams, and other customer or employee communication and knowledge sources.

The Ask-AI platform collects data, analyzes it and shares insights with company executives. Executives tout the technology as a tool that supports employees’ workflow and helps them be more productive, providing multiple use cases for customer support, success, product, research and development, and sales. However, there are limits to what employee data will be shared.

Alon Talmor, founder and CEO of Ask-AI, said the tool doesn’t connect to sensitive systems such as payroll applications, because of the nature of the data. However, the tool does retrieve and examine other key data points that are generated in systems with less-sensitive data as employees work through their tasks.

“What we do connect with is company portals in which there’s a lot of information about what an employee is entitled to, what are the onboarding procedures and any kind of processes the company follows,” Talmor said.  

Eventually, Talmor would like to break down more siloed repositories of employee data, but he understands that it’s a slow process.

“Our strategy is to start with customer-facing teams like support success, then we move to sales, then we expand to the rest of the organization with applications that are relevant to that job role. We are still early with the HR applications, but it is definitely something we’d like to expand on. We’d like to build specific HR applications that would help with HR’s day-to-day tasks,” Talmor said.

Nicole Lewis is a freelance journalist based in Miami.