Rob Green is the chief digital officer of Insight Enterprises, a Fortune 500 Solutions Integrator.getty​AI is now moving firmly out of its hype cycle and into practical applications at scale. Agentic capabilities are expanding, and agent-to-agent coordination is now accelerating workflows and improving productivity. The practical use cases are expansive and can be applied across many enterprise functions and flows. Agents are now handling intricate customer service tasks, writing complex code and increasingly serving as co-workers who take care of mundane tasks.These are just a few of the practical applications we’re increasingly seeing with AI after setting out a year ago to help our internal lines of business identify AI use cases ourselves. At Insight, we've leaned into AI to drive internal awareness and adoption to get teammates to embrace AI’s potential. To support their AI journeys, we’ve introduced agents to help reduce low-value, add-task work and drive higher-value client interactions.What we learned: • Low-hanging use cases mean quick wins that are key to buy-in.• Quick losses can be just as critical to long-term success.Ultimately, companies shouldn’t be afraid to fail with AI, especially if the losses come out of the starting gate. Not every AI deployment is going to be a home run, but learning quickly by assessing business impact, monitoring adoption and pivoting when the investment doesn’t deliver the expected outcomes is critical.We experimented early on with AI initiatives to improve time to serve clients and reduce workflow complexity, which missed the mark. In the process, we learned a lot about the limitations and capabilities of AI. That led us to create a platform and methodology to validate ideas, put them in the context of business outcomes and then prioritize them to realize ROI faster.Another big win came in the form of our internal AI training platform, which helped gamify upskilling, so teammates could incorporate the technology into their day-to-day and become more productive. In each case, we rolled the platforms out internally before determining how useful they could be at scaling adoption for other companies.Our success came down to three pillars.1. Architectural Freedom To Explore Alternatives To Legacy InfrastructureThese platforms were new builds, and that's important. We discovered we can accelerate development when we're not saddled with legacy infrastructure and code. In fact, we leaned into using AI to help generate code, which improved speed to market, especially when it came to agents.Take our e-commerce platform, for example. It was hard for an agent to sift through legacy environments and figure out how to write useful code. Starting fresh, without trying to put legacy code into context, AI is better equipped to deliver impactful results. That’s the tack we also took, developing an agentic B2B buying agent to support our clients and help them identify optimal solutions for business-critical use cases. Overall, AI wrote 70% of that code, helping us put out an early version to test in three weeks. It would have taken six months had we done it manually.Legacy infrastructure still played a role, especially in the agent’s internal validation by our dedicated development team. The app itself was also thoroughly tested by teammates and clients of ours during a rigorous beta phase. In the end, many foundational elements contributed to getting it to market, but, in an era where technology is constantly changing, it’s necessary to find a balance between what’s tried and true and what’s transformative. 2. Strategic Partnerships To Validate Use CasesThe goal with our buying agent is to help take our clients’ B2B procurement to the next level. Imagine a world in which B2B buyers can ask for relevant solutions with a natural language interface: “I'm a small school district in central Iowa with 2,000 students, and I need to get laptops to my teachers. What do you recommend?” Our agent is capable of recommending optimal solutions that can be transacted instantly online.Agentic buying works so well because you can train it on purchase history, client preferences and contextually relevant solutions. That enables our agent to propose solutions that specifically match client needs. Future client interactions will help further streamline the process, as it learns more and more. That’s just one way our clients contribute to our success. Both clients and partners can also take more of an active role in getting AI tools to the developmental finish line, offering trusted feedback through official channels. Strong relationships with customers and partners are invaluable to the development and even ideation of solutions. Innovation doesn't happen in a vacuum.3. Enterprise-Wide Adoption Starts At The TopBefore even a single line of code is written or generated, leadership has to firmly commit to a project. The quick wins to get your feet wet with AI are one thing. However, with nearly two-thirds of respondents to McKinsey & Company’s “The state of AI in 2025” survey saying they haven’t yet begun scaling AI across their companies, the kinds of platforms I’ve mentioned clearly represent radically different animals.The larger the initiative, the greater the buy-in at the top of the organization, and that’s just to gain traction. To get a project off the ground, you need a cross-functional team. And to find success, you need everyone in the organization moving in the same direction. That only starts with leading by example. ​​Based on recent PwC research, people, including senior executives, can hold back the development of AI agents. The top-ranked reason among the 18% of respondents that weren’t using agents (at all) was a lack of clear use cases. However, based on our experience, if you’re looking, use cases can easily be found. And, with our training platform, where all teammates—including our CEO—incrementally gain different levels of AI expertise, people are also the solution. That’s the only way to move past the hype and move forward in the cycle within your own organization and accelerate agentic development: to first simplify AI. There’s no denying its complexity. There's only embracing it before you can start to apply it practically.Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?