AI strategy can be built on sand, and in this case, it’s a good thing

Growing up the sandbox was full of treasure: weird shaped rocks, pillbugs, coins, tiny plastic toys left and forgotten by a previous visitor. It’s a place of wonder for children, provided the neighborhood alley cat doesn’t also bury treasure there.

Sandboxes are small plots dedicated to play, creation, and sometimes destruction, especially in service of building something greater. It’s an apt analogy for a space created for employees to play with AI at work.

Rather than rely solely on IT departments or outside consultants to implement AI tools across the organization, some companies are creating internal ‘sandboxes’ of sorts that provide employees with the space to experiment, tinker, and even build their own applications, and are finding it’s a great way to center an AI strategy.

Thinking about AI from the ground up rather than top down helps with adoption inside the organization. It can ease anxieties about AI’s place at work and how humans fit into the new paradigm. It also can further innovation from inside the company, with the aid of the people who actually do the work: your employees.

“Organizations, if they want their company to survive through this transformation and thrive…they need to be providing employees the opportunity to use these tools and understand how they can evolve, how it will evolve business models,” said Mary Alice Vuicic, CPO at Thomson Reuters.

Playpen. Vuicic and her colleague from the technology team recognized the transformation potential early after the launch of ChatGPT, and set out to articulate a strategy focusing not only on the global news and professional services firm’s approach to adopting the technology but also its talent strategy amid this AI transformation.

At Thomson Reuters, the people strategy centered around “Four Ts” to prepare its workforce for AI: tone from the top, training, tools, and time.

“You have to get hands-on experience to understand,” she said of “tools” and “time.” “We’ve seen a huge shift when that happens.”

Thomson Reuters first launched Open Arena in 2023. The secure internal “sandbox” offers the entire Thomson Reuters workforce, all 27,000 employees, access to leading large language models to explore AI in the workplace and gain hands-on experience interacting with the tech to foster a culture of experimentation, as well as trial and error.

“Everybody can use it, and it is secure,” Vuicic said. “We encourage people to go in [and] start their work in that sandbox.”

More than half of Thomson Reuters employees are now using Open Arena regularly, and the company is starting to track usage and report its findings back to employees and their managers to foster transparency about AI use at the company.

Open Arena also helped facilitate the company’s AI champions program. Some 400 Thomson Reuters employees are AI super-users and are helping to coach peers and design use cases on how to better use it at work, she said.

“Innovation and problem solving can come from anywhere in the organization,” Vuicic said. “That’s very different from how a traditional organization structure works.”

Homegrown experts. Other global companies are also seeing that employee involvement is critical to scaling AI. At collaboration and team-based project management software company Atlassian, non-technical employees aren’t just end-users, they’re also AI builders.

Atlassian uses its customer-facing generative AI tool, Rovo, internally. Atlassian’s strategy includes engaging teams across the company to build tools tailored specifically to their workflows.

CPO Avani Prabhakar told HR Brew that this results in both faster adoption and deeper ownership over how AI shapes work for Atlassian employees, pointing to widely used agents built out by her HR team.

“We have Rovo agent, that’s the tool that you get,” Prabhakar said. “Our team has been able to build an onboarding agent on top of it. That agent is called Nora…It is built by [the] HR team, so I don’t have to worry about adoption.”

Because HR leaders involved in onboarding at Atlassian helped ideate how AI could improve or impact the experience for new hires, Nora acts more like an “onboarding buddy” for new hires, answering “dumb” questions like those a new hire may be too embarrassed to ask a supervisor or colleague and to walk employees through their early days. This tool, built by actual HR professionals, is used by 70% of new hires, according to Prabhakar.

“The good thing with non-tech users like HR is like, ‘Oh, I can create my own agent, it’s not being done by the IT team and then given to me to work with,’” she said. “ It’s just a very different notion.”

Atlassian’s most popular AI agent lives inside its performance management system, APEX, Prabhakar adds. It can assist in drafting talking points, and also help employees create written content or feedback. The tool also helps managers consolidate information from 360-degree feedback, peer feedback, and other places into a coherent assessment.

“Because you are letting employees tinker with the technology, then there is more buy-in, but also it eradicates all the fear that sits around it,” Prabhakar said.

Vuicic and Prabhakar agree that letting workers explore the tools for themselves—and not simply participate in training—creates a shift in mindset and eases adoption and fear. Adopting the new technology becomes less about enforcement and box checking and more about learning, engagement, and career transformation.

“They get past anxiety or fear about, ‘What is this technology’ and, ‘What’s it going to do to my job,’” Vuicic said. “But then it becomes a delight, because it’s so accessible,” adding that she imagines a future wherein the remaining work—what’s not modified or impacted by AI—actually becomes more human.

And that unexpected employee delight is a business advantage too, Vuicic said, adding that top talent expects access to AI now.

“They want to be future-proofed, and so they will leave and go to companies, or they won’t join companies that don’t give them that access,” she said.