For HR leaders, getting started with AI is about more than choosing a use case. It’s about adoption. How do you help people feel comfortable building with it? How do you make them confident enough to use it?
Spectrum provides internet, mobile, TV, and phone services to customers in 45 states. Its HR technology team has been exploring how to build confidence in AI as it develops new capabilities. During a recent conversation, Derek Dreyfus, vice president of HR technology, shared how experimentation, early users, and ongoing support shaped his team’s experience. Here are five lessons they learned while building their own AI team and developing agents in Oracle AI Agent Studio.
1. Give people room to experiment
Spectrum began with lunch-and-learns, giving colleagues space to experiment, test ideas, and discover what AI Agent Studio could do. That curiosity soon carried beyond the sessions, with team members continuing to experiment independently. Some of those colleagues went on to join Dreyfus’s team, developing their skills through hands-on work and Oracle training. For Dreyfus, AI offered an opportunity to “grow as a team, grow as individuals.” Involving people early helped reveal who wanted to take that next step.
2. Make the first experience worth repeating
“Start with a useful problem,” Dreyfus advises. Rather than jumping into complex agentic applications, he recommends building smaller agents first, giving teams time to develop their skills and users a chance to see the benefits firsthand. A disappointing first encounter can be difficult to overcome. Even after the technology improves, people may be reluctant to give it another try. Starting small helps teams build confidence and lays the groundwork for more ambitious AI applications. But the goal isn’t simply to launch an agent. It’s to solve a problem that matters to colleagues and give them a positive experience that makes them want to use AI again.
3. Let early users become AI advocates
Spectrum started by introducing agents to pilot groups in selected business units, giving employees an opportunity to experience the technology firsthand. The initial testers were enthusiastic about participating, and that excitement extended beyond the pilots. They began sharing their experiences with colleagues in other parts of the company, becoming advocates for the agents they had tested. “It’s socialized organically, which is great,” Dreyfus says. Giving people the freedom to form their own opinions about AI can be more compelling than asking them to embrace something they haven’t tried.
4. Own what happens after launch
For Dreyfus, making AI a full-time responsibility was about more than building agents. It meant Spectrum’s HR technology team could develop and improve capabilities without waiting in another team’s development queue. “We control our path,” he says. That ownership extends well beyond launch. As usage has grown, Dreyfus says the team spends hours reviewing questions from the previous day, identifying needs they hadn’t anticipated, and refining agents accordingly. The team now releases an estimated three to four AI Agent Studio enhancements a week, including new agents and improvements based on feedback. Working closely with legal and compliance, the team can continue improving the experience while maintaining appropriate oversight. Rather than treating the launch as the finish line, Spectrum has made ongoing improvement part of the team’s responsibility.
5. Build expertise without going it alone
Spectrum wanted to own its AI development, but Dreyfus knew his team would need support along the way. Early on, Spectrum approached Oracle for help building its confidence and capabilities. That relationship has continued through weekly meetings with Oracle’s AI team, where they troubleshoot technical challenges, discuss upcoming features, and share feedback on areas where the technology could improve. The collaboration goes both ways. Spectrum gets help resolving issues, such as connecting third-party applications and navigating authorization requirements, while Oracle gains insight into how its technology performs in practice. Dreyfus credits that additional support with helping Spectrum move faster and sees working through challenges together as a valuable learning experience for both teams.
Confidence built through experience
Spectrum’s experience points to a practical path: give people room to learn, make early experiences worthwhile, and give the team behind the technology the ownership and support to keep improving it. Confidence grows as colleagues discover what AI can do for them, share what they’ve learned, and help shape what comes next.
