How Agile Enterprises Are Re‑Thinking AI Strategy – Lessons From Freshworks and Dennis Woodside

When you ask a tech leader what the biggest challenge of 2026 is, most will point to the relentless pace of AI. It’s not just about building cool models anymore; it’s about wiring those models into the day‑to‑day rhythm of a company that prides itself on being agile.

Freshworks, the customer‑engagement platform that grew from a scrappy startup into a $6 billion public company, has found a surprisingly simple answer. In a candid interview with Fortune, its chief executive, Dennis Woodside, explained how the firm is treating AI as a series of short, iterative experiments rather than a monolithic, once‑and‑done project.

“We stopped thinking of AI as a ‘big‑bang’ rollout,” Woodside said, chuckling. “Instead, we ask the same question we always ask for any new feature: Can we ship something useful in two weeks, learn from it, and then decide whether to double‑down?” That mindset, he argues, is the secret sauce that lets a company that employs roughly 4,000 people move at the speed of a startup while still maintaining enterprise‑grade reliability.

The first step, according to Woodside, is to identify low‑friction use cases where AI can deliver an immediate lift. For Freshworks, that meant augmenting its ticket‑routing engine with a generative‑text model that could suggest the most relevant knowledge‑base article in real time. The result was a 12 % reduction in average handling time—a metric that, while modest on paper, translated into thousands of saved support hours each quarter.

What’s crucial here is the modesty of the ambition. “If you aim for a moonshot and miss, you waste time, money, and morale,” Woodside warned. By starting with a clear, measurable win, the team could prove value quickly, win over skeptics, and, importantly, build a data set that would fuel the next round of improvements.

Once a quick win is in the bank, the next phase is to broaden the scope—always with a tight feedback loop. Freshworks set up a “AI guild,” a cross‑functional group of engineers, product managers, and compliance officers that meets twice a week. The guild’s charter isn’t to approve every model; it’s to surface friction points—bias concerns, data‑privacy gaps, latency issues—and to iterate on solutions before they reach customers.

This governance structure, Woodside emphasized, is deliberately lightweight. “We don’t want a bureaucracy that slows us down. The guild is more like a standing coffee‑break discussion where we flag red flags early, not a committee that signs off months later.” In practice, the guild has helped Freshworks avoid a few embarrassing AI misfires, such as a language‑generation bug that unintentionally suggested pricing changes in a support reply.

Scaling the approach across the organization required a cultural shift, too. Freshworks rolled out an internal “AI literacy” program that blends short video modules with hands‑on labs. The goal isn’t to turn every salesperson into a data scientist; it’s to give employees enough context to ask the right questions and recognize when a model’s suggestion might be off.

That cultural layer is where Woodside sees the biggest differentiation among so‑called “agile enterprises.” “If you have the tech but not the mindset, you’ll end up with a bunch of models that sit on a shelf,” he said. Conversely, a workforce that treats AI as a teammate—questioning, tweaking, and sometimes overriding it—creates a feedback‑rich environment that drives continuous improvement.

Of course, the financial side matters. Freshworks set an internal KPI that the ROI of any AI initiative must be evident within a six‑month horizon. This rule keeps teams honest and forces them to think about cost, not just capability. The ticket‑routing experiment, for example, paid for itself in less than four months when you factor in reduced labor costs and higher customer satisfaction scores.

Looking ahead, Woodside admits the company is still early in its AI journey. The next big push is a set of generative‑AI assistants that help sales reps draft personalized outreach emails based on a prospect’s recent activity. The plan mirrors the earlier rollout: prototype in a two‑week sprint, test with a pilot group, measure lift, and then decide whether to scale.

What’s refreshing about this roadmap is its humility. Instead of betting the farm on a single, polished AI product, Freshworks embraces a series of small, test‑and‑learn cycles—a philosophy that aligns neatly with its broader agile development culture.

For other enterprises watching from the sidelines, the takeaway is simple: treat AI like any other feature you ship. Start small, iterate fast, embed governance as a conversation rather than a gate, and make sure the whole organization can speak a little AI. If you can do that, you’ll likely stay ahead of the curve without getting tripped up by the inevitable hiccups along the way.