Can Gen AI Help Agile Teams Improve Delivery Clarity?

Not every AI story begins with a big launch. Sometimes, it begins with a report that takes too long to explain. A spreadsheet that is already out of date. A team trying to fix the same issue again next week.

For me, that is where the journey began.

I’m a manager of agile program management. My role places me close to the everyday rhythm of delivery: tracking progress, helping teams stay aligned, supporting product conversations and making sure work moves forward with clarity.

In that role, I saw teams doing what good teams do. They were tracking progress, sharing updates and solving problems. But many updates still depended on reports, meetings and manual follow-ups. By the time an issue became visible, the team was often already in fix mode.

I started asking a practical question: How can we foresee (and prevent) these issues?

My interest in data helped me explore that question. I liked looking beyond the numbers in a report to understand what they meant. Why was something delayed? What pattern was repeating? What could the team learn before the next meeting?

That led me to dashboards.

At first, the work was simple. Bring the right data together. Make it easier to read. Help teams see risks, delays and patterns in one place. Some people needed time to see the value. A dashboard can look like just another chart until it starts changing the quality of a conversation.

Slowly, those conversations changed. Instead of relying only on manual updates, teams could look at the same information and discuss risks, delays and recurring patterns from a shared starting point.

But visibility was only the first step. The dashboards could show us where problems were occurring; they could not always help us prevent the same problems from returning.

That became especially clear when we looked at recurring quality issues in work items, including unclear goals, incomplete descriptions, limited detail about target users and uncertainty about the intended outcome.

I began asking a new question: Once the data helps us identify a recurring problem, could AI help us address it earlier?

That question connected my interest in data with my growing interest in gen AI.

I did not approach AI as a separate initiative or as something to use for its own sake. I looked at a problem the team could already see and considered whether an approved AI tool could help improve the quality of the work before it moved further through delivery.

The pattern was familiar. Issues were identified, teams fixed them and then similar issues appeared again later in work items. The challenge was not simply formatting. It was whether the work clearly described the goal, the target user and what the team needed to deliver.

This created an opportunity to test whether gen AI could help improve clarify earlier in the process.

I worked with the product owner to test an approved gen AI tool on a single epic. My goal was simple: keep the meaning and intent intact, but make the content clearer and easier to act on.

Because the tool was approved and I used it on one real work item, my team could review the output carefully, validate the meaning and make sure AI improved clarity without changing the intent.

The result was encouraging. In that example, the number of quality issues came down from around nine or ten to one.

It was a limited experiment, but it showed us something useful: data could help reveal a recurring delivery problem, and gen AI—used with human context and review—could help address that problem earlier.

AI was helping us improve quality, save effort and think about issues earlier. It showed me how a tool, when used with context and care, can support better delivery.

Environment matters.

I had leaders who encouraged me to try available tools. I had product partners who were open to testing a new approach. I had space to learn through certifications, webinars, articles and hands-on practice. I also had forums where I could share what I was learning with others.

That support matters in AI work.

People do their best learning when they are close to real problems. We grow when we can ask questions, test ideas and learn from strong teams. We build confidence when experimentation is supported, but also grounded in client value.

My path into AI was not a straight line. I came from a quality engineering background. Data became an area of interest through real project work, and AI became another step in that learning journey.

My advice to others is simple: start with curiosity. Read a little. Attend a session. Ask someone who knows more. Try one tool on one real problem. Then share what you learn. You don’t need to begin with a large AI program. Starting with one clearly defined problem can make it easier to learn what the technology can do, where it has limitations and where human judgment remains essential.

For AI talent, that is what makes work at Publicis Sapient meaningful. The challenges are real. The teams are collaborative. Learning is continuous. And the impact is connected to work that clients and teams can actually use.

Because the future of AI is not only built through large breakthroughs. Often, it starts with someone noticing a problem, asking a better question and being given the space to try.