Key takeaways
- Start with small, low-risk use cases to build organizational trust before tackling complex AI implementations
- Measure AI impact across multiple dimensions — efficiency, transformation, and risk — not just hours saved
- Implement governance frameworks early to manage agent sprawl as adoption scales from dozens to thousands of agents
Projects stalled in the pilot. Poor adoption numbers. Unwelcome surprises related to costs.
You know the story. Businesses continue to struggle with AI implementations, and you’re likely feeling the same pain.
AI is moving so fast, and organizations understandably fear falling behind. But they also don’t know how to get started. It’s why so many are spinning their wheels and are unable to show measurable value.
This harsh reality is why we created the Agentic Impact Workshop series.
We’ve conducted more than 250 structured sessions, and counting. We’ve worked with enterprises of all sizes across retail, manufacturing, healthcare, higher education, and other categories. These are day-long co-innovation workshops where we help them take their first steps on the AI journey by focusing on practical projects that deliver real-world value. Our goals are straightforward. Show what’s possible with AI. Collaborate on ideas that make sense for their businesses. Then, work together to build production-grade prototypes.
Our sessions with both IT and line-of-business leaders have identified more than 1,100 use cases, demonstrating the strong appetite to embed AI-driven automation into processes. We’ve also learned some important lessons about why organizations get stuck in the early days of AI. While organizations may vary by industry and have vastly different goals, we consistently observe four common issues that any business hoping to get AI off the ground must address.
- Building Trust
- Measuring Impact
- Showing the “Aha” Moment
- Maintaining Governance
Recently, I’ve presented what we learned in front of packed sessions at HumanX and AWS Summit New York City. So I know firsthand there’s an incredible thirst to learn how to successfully launch AI initiatives.
As I delve into these more closely, you’ll likely recognize gaps that are holding you back in your projects.
Steps to Building Trust
AI is a leap of faith. We’re trusting our businesses with a new technology that only burst onto the scene in late 2022. How do you trust the decisions it makes? How do you trust IT to implement it correctly and the business to use it responsibly? And perhaps most important of all: can you trust the quality of the data that provides the business context for AI models and agents?
If you’re being honest, your data probably isn’t AI-ready. Data in our enterprises has never been truly ready for anything. I’ve spent a lot of my career in the business intelligence space at companies like SAP and Microsoft, and the data foundation is always the most difficult thing to wrangle. Now, with AI, it’s even more essential. That’s because data is no longer just for humans. Agents are also reasoning with the data.
AI also adds to the inherent tensions that have always existed between IT and the lines of business. IT is frustrated by the use of shadow AI across the organization. The business is frustrated because IT isn’t moving fast enough. Both are feeling the pressure from the board and the CEO to go faster, be more productive, and do more with less. So establishing trust between the teams is essential to making AI successful.
The best way to address trust is to start at the bottom. What we constantly hear in our workshops is the desire by businesses to dive into the deep end of the pool with highly complex use cases. We recommend that you wade in slowly, beginning by just making the mundane better. Low-risk use cases play a very important role in building trust. You’re showing wins that give you the confidence to move toward those more complex use cases.
Measuring Impact Dilemma
If you can’t measure it, you can’t improve it. And if you can’t tie value to AI, then it’s just an expensive science project. Traditionally, impact metrics were simple: hours saved or dollars generated. That applies to AI, too. But associating that kind of value has proven problematic. In the eagerness to go from pilot to production, people don’t take the time to define what they want to achieve and then decide how to measure it.
Also, it’s important to consider identifying measurements that aren’t readily found on a balance sheet. For instance, what does success look like in terms of your business’s organizational transformation?
Early on, you may have an automation use case where you can measure the efficiency gains of moving data from Point A to Point B. But as you move along the maturity curve into more complex use cases, you’ll want to tell agents to start acting and fixing things within the business. Your measurement metrics need to adapt to that reality. For instance, in sales, you might have an agent that essentially is part of the team and can help prevent deals from going south. That will require a different way of gauging impact as you transform the organization into a more agentic one.
Then there’s risk management. Are the agents creating any unexpected vulnerabilities? When you add an agent to a process that involves employee access to Salesforce, it might change what people can view. Maybe they can suddenly see all of the revenue and pipeline, which wasn’t what you intended. Observability into what agents are doing is crucial because you don’t want AI to be a black box. You need to know why agents made their decisions and whether they were the right ones.
Finding Sparks That Drive Change
We explored earlier the idea of small wins that drive big changes. Businesses need to show progress around AI. It’s more than simply justifying the time, money, and resources invested. Right now, people are still sifting through all the claims about AI, trying to determine what’s real and what’s hype. Many employees are worried about the impact on their jobs.
Success flips that narrative. Wins get noticed. When everyone in the business sees productivity gains, that builds momentum throughout the organization. People start to see the bottom-line value of AI, and the positive feelings begin to resonate. Businesses feel better about the investment. Employees see how AI makes them more effective at their jobs, not more likely to be replaced.
Providing this spark acts as a catalyst, helping people understand how it actually adds value to the organization. This is where progress becomes real.
Preparing for Agent Sprawl
A recent Gartner report, “Beyond Agent Sprawl: The Rise of AI Agent Management Platforms,” noted that the average Fortune 500 enterprise used 15 agents in 2025. By 2028, that number is expected to explode to over 150,000 agents 1. Agent sprawl is real. Governance is essential. As the development of agents becomes commoditized and they’re found across every part of the business, you need visibility into anomaly detection, guardrails on what they can and cannot do, ongoing cost tracking, and a way to ensure you never lose control of their actions.
Our solution to this growing challenge is Boomi Agentstudio. It gives you a single cockpit to manage all of these agents and detect issues when they’re surfaced for swift action. And if the agents aren’t making the right decisions, you can then adjust by digging into the data sources that the agents are relying upon.
The Art of the Possible
So, here are some quick takeaways:
- Start small before laddering up to more complex use cases
- Build a culture of trust around AI within the organization
- Measure value across dimensions you might not normally consider
- Create breakthrough moments that make everyone notice
- Bake in governance from ground zero
Our workshops are designed to bridge the fundamental disconnect between the mounting pressure to use AI and the need to understand why you want to do something. Before you build agents, you first need to build a trusted foundation to support them. That’s when agentic transformation happens.
AI agents are contagious. Once organizations get going with agents, they thirst for more. Agentic processes start to snowball. Businesses have the ambition, but sometimes they just need a little direction.
1 Gartner report, “Beyond Agent Sprawl: The Rise of AI Agent Management Platforms,” Anushree Verma, March 6. 2026. GARTNER is a trademark of Gartner, Inc. and/or its affiliates.