AI agents join your team: 3 rules from early adopters

Key Takeaways

• Gartner projects AI agent software spending will hit $206.5 billion in 2027, up from $86.4 billion in 2025
• Early adopters report AI agents free up human time from routine reporting, enabling more strategic work
• Success requires continuous benchmarking, openness to tool changes, and focusing agents on the right tasks

AI agents are entering production environments faster than most enterprises anticipated. At the Snowflake Summit 2026 in San Francisco, three digital leaders from Fanatics, Whoop, and other organizations shared hard-won lessons about making these agentic colleagues deliver real value. Their advice centers on three areas: benchmark relentlessly, stay ready to swap tools, and point agents at the work humans don't want to do anyway.

Gartner's numbers make the scale clear. AI agent software spending is projected to reach $206.5 billion in 2027, more than double the $86.4 billion expected in 2025. By 2028, the analyst firm predicts 33% of enterprise software will include agentic AI capabilities. Companies are betting heavily that agents can handle tasks from basic operations to autonomous decision-making.

Why benchmarking AI agents matters more than you think

Madeleine Want, VP of data at Fanatics, runs a rigorous tracking program across her data practitioner community. Her team doesn't just deploy agents and hope for the best. They measure how people use these tools, what tasks they apply them to, how much time they save, and what they do with the recovered hours.

"Every business analyst out there will tell you some version of, 'I wish I could be doing more strategic work, but I am bogged down in routine reporting,'" Want said. "What we are seeing is that the more routine reporting tasks are the ones that often lend themselves best to automation through AI."

The benchmarks confirm what many suspected: agents excel at repetitive work, freeing humans for judgment calls. Staff get time back and redirect it toward strategic work. That's the outcome executives hope for, but Want's approach shows you need data to prove it's actually happening.

Hold your tools lightly

Want offered a counterintuitive warning: don't get too attached to any particular agent or platform. Fanatics tests tools constantly, compares features, runs previews, and builds design partnerships with vendors. The goal is staying ahead of a fast-moving market, not locking into today's best option.

“This is not your traditional enterprise technology multi-year transformation project. We are not adopting well-tested, well-trodden technologies that, once rolled out, will never be rolled back. We're in an experimental phase right now.”

— Madeleine Want, VP of Data, Fanatics

Her advice: adopt early, try things, but stay agile. The agent that works today might be obsolete in six months. Building organizational muscle for rapid tool transitions matters more than picking the perfect solution right now.

Where to point your AI agents first

Matt Luizzi, VP of analytics at Whoop, was already focused on freeing his team from low-value work before agentic AI arrived. The technology accelerated his existing strategy. Both leaders converge on the same principle: agents should handle tasks that humans find tedious and repetitive.

This isn't about replacing people. It's about reassignment. Agents take on routine reporting, data aggregation, and operational tasks. Humans focus on interpretation, strategy, and decisions requiring context that agents can't access. The split works when companies are explicit about which work goes where.

Microsoft's Work Trend Index research suggests a 40% productivity improvement for knowledge workers using AI agents on routine tasks. But that improvement only materializes when organizations deliberately route the right work to the right worker, whether human or agent.

What happens when agents fail

Gartner estimates 40% of enterprises will scrap their AI agent initiatives. The failures usually trace back to unclear expectations, poor task selection, or treating agents like traditional software deployments. Agents require ongoing management, not set-and-forget implementation.

Want's back-and-forth process between managers and professionals reflects this reality. Deploying agents means continuous discovery of new AI-enabled workflows, not a one-time rollout. Organizations that expect transformation project timelines will likely end up in that 40%.

Frequently Asked Questions

How much will companies spend on AI agents by 2027?

Gartner projects AI agent software spending will reach $206.5 billion in 2027, up from $86.4 billion in 2025.

What tasks are AI agents best suited for in the workplace?

Routine reporting, data aggregation, and repetitive operational tasks. Early adopters report agents excel at work humans find tedious, freeing staff for strategic responsibilities.

Why do 40% of enterprise AI agent projects fail?

Common causes include unclear expectations, poor task selection, and treating agents like traditional software with set-and-forget deployments. Agents require ongoing management and continuous workflow discovery.

How should companies measure AI agent success?

Track how tools are used, what tasks they're applied to, time saved, and what employees do with recovered hours. Self-reported value metrics help quantify whether agents deliver real productivity gains.

Source: Latest news