When people talk about AI, they usually focus on speed, automation, and efficiency. But there’s a quieter, more human story unfolding inside product teams who are actually building with it.

I sat down with Maggie La Belle, a product manager with experience at Ada and Zendesk, to dig into what it really means to build customer experiences that scale.

With a background in linguistics and a career shaped by conversational systems, Maggie brings a unique perspective: product work isn’t about replacing people—it’s about designing for trust, nuance, and real-world complexity.

This interview explores how her work has evolved, what’s changed (and what hasn’t), and why empathy is still the sharpest tool in a PM’s toolbox.

From speech therapy to product management

Maggie never set out to be a product manager. She planned to become a speech therapist, but her fascination with systems and communication led her down a different path into tech. What stuck with her was the idea that both language and product are structured systems—fluid, layered, and meant to be understood and shaped.

That background gave her an edge when she entered the world of messaging platforms. From an early-stage healthcare startup in Montreal to working on business messaging at Zendesk, she’s always been drawn to building systems that help people communicate at scale.

“Language is a system. Product is a system. That’s what drew me in.”

The shift to generative messaging in customer service

When Maggie joined Ada, she stepped into a company already deeply invested in conversational systems. Ada had long focused on natural language processing to power its customer support agents. But recent breakthroughs in language models changed the game.

Today, Ada powers the first line of communication between customers and companies like Pinterest and Square. The user might start a chat on WhatsApp or a mobile app, and interact first with Ada’s AI agent before escalating to a human—all in the same thread. That seamlessness is the real innovation.

Behind the scenes, Maggie and her team were balancing the needs of customers adopting a modern conversational AI agent while continuing to support the full spectrum of technologies that had evolved over time. With multiple generations of capabilities running within the same system, product decisions often came down to tough trade-offs: How do you prioritize feedback across different customer journeys? When do you reinforce what’s already working versus invest in where the product is going?

“It’s the same platform, but supports totally different implementations running on top of it. That’s a real product challenge.”

Product managers working in evolving architectures need strong prioritization frameworks.

• Defining which user segments are still dependent on the legacy path—and for how long.
• Flagging any feedback that could steer both systems in a better direction.
• Committing to a clear migration strategy instead of endlessly supporting both paths.

For teams dealing with platform transitions, it’s a reminder: legacy systems shouldn’t dictate your roadmap, but they should inform your transition playbook.

Building products requires more empathy, not less

Automation might change the interface, but building great products still demands deep human understanding.

Maggie shared how some of the hardest work is helping customers reframe what they expect from a chatbot. The instinct is to control every step. But conversational interfaces don’t follow rigid flows. That introduces risk—and anxiety.

“Our job is to help customers trust the system, but also to support when they to lock it down for certain business policies or service flows. It’s not one-size-fits-all.”

Product teams now have to answer: when is it safe to allow flexible interactions? When is it essential to stay deterministic? And how do you create systems that adapt across industries with radically different levels of maturity?

For Maggie, this comes down to coaching and being a true consultant of the product. Customers need to retrain how they write help articles, how they structure guidance, and even how they evaluate quality.

“You’re not just building a product. You’re helping someone else rewire how they think about customer support.”

Feedback is still manual, still messy

Even with more automation in the loop, insights don’t magically fall from the sky. For Maggie, they still come from Slack threads, Gong calls, and internal questions from CSMs and AEs.

“Sometimes it really is just a bug. But often, it’s about how the customer is framing the problem. That’s the real insight.”

She tracks where the request came from, how the CSM phrased it, and what that tells her about the real friction. And then she cross-references that with Ada’s roadmap: is this something that fits into a longer-term platform improvement? Or a one-off fix?

Cycle and Glean are the tools that help her manage scale.

With Glean, Maggie performs deep-impact research across scattered systems like Slack, Notion, Jira, and GitHub. If she’s digging into a bug or feature request, she can quickly surface past threads, documentation, and internal discussions that help her understand the root cause—or how the team has previously tried to solve it.

Cycle helps her trace feedback to its origin. Whether it’s a quote from a Gong call, a Slack message from a CSM, or a customer ticket, she can cluster those inputs together, connect them to roadmap features, and prioritize the most common or high-impact themes. It makes the invisible visible.

But she’s clear: these tools don’t replace product judgment. They widen the lens—so she sees more, faster—without drowning in noise. They don’t replace judgment. They augment it.

What tools can (and can’t) do for PMs

Search, summarization, and documentation support? Yes. Product strategy and decision-making? Still on the PM.

For Maggie, tools like Cycle and Glean are most powerful when used to expand visibility—not to replace thinking. The key is knowing what question you’re trying to answer. If she’s investigating the impact of a customer issue, she starts with Glean to explore historical conversations across Slack, Jira, Notion, and GitHub. If she's clustering related feedback across teams, she uses Cycle to follow the thread from source to feature.

But tools only do their job when you stay intentional. Maggie doesn’t let AI or automation write the core documents that define her product's direction. Docs, one-pagers, and scoping work are still where she embeds context, nuance, and decisions.

“I haven’t found anything that sounds like me in long-form. So if it’s a one-pager, I write it myself.”

Her advice to new PMs?

• Shadow support. Watch how bugs are handled. Reproduce them yourself.
• Learn your product by actually using it.
• Don’t skip the messy parts. That’s where the truth is.

“If you come in thinking you can shortcut your way through product management, you’re going to stay surface-level.”

At the end of the day, tools can be multipliers— but only if you bring the product sense, the discipline, and the empathy to guide them.

‍