Ameren
Designing a conversational AI interface that turned fragmented policy documentation into instant answers — reducing support call time by 63%.
Athena AI
Help desk employees were spending more time searching for answers than giving them. We fixed that — and cut average call time by 63%.
Deep Dive
Background
Ameren is a Fortune 500 regulated energy utility and the largest electric provider in Missouri, serving 2.4 million customers across Missouri and Illinois. Their commercial customer base — businesses, municipalities, and large facilities — requires specialized support that is more complex and time-intensive than residential service. That complexity was creating a serious operational problem for the help desk.
Role
Lead Product Designer, working alongside a product owner, 2 product managers and an engineering team.
The Problem
Commercial customer support calls at Ameren were taking an average of 32 minutes to resolve.
The wait times that created were degrading the experience for every commercial customer in the queue.
When we started digging into why calls were taking that long, the answer was not what anyone expected…
Research
I interviewed help desk employees directly and observed them as they processed live tickets — watching the actual workflow unfold in real time.
We layered in platform analytics to understand how employees were navigating the support tools, and sent surveys to commercial customers who had recently submitted tickets to capture the experience from their side.
What We Found
Help desk employees were not slow because the problems were hard. They were slow because finding the right information was hard.
Policy documentation, rate structures, billing rules, and service procedures were scattered across internal systems with no unified way to surface the right answer quickly.
Employees spent the majority of each call searching — not helping. The average handle time of 32 minutes was largely search time in disguise.
The Realization
The solution was not a better search tool. It was eliminating the search entirely.
A conversational AI interface trained on the complete body of Ameren policy and procedures could put the right answer directly in front of an employee the moment they needed it — in plain language, with no manual navigation required.
What I Designed
Athena AI, a conversational AI tool, trained on Ameren's full policy documentation and built to be as frictionless as a chat interface.
The UX work centered on the most consequential design decisions in any human-AI tool: how the system communicates confidence levels, how it handles edge cases and policy gaps, how it surfaces source context so employees can verify answers when stakes are high, and how it escalates gracefully when a query falls outside its training.
The interface was designed for help desk employees under pressure on live calls — which meant speed, clarity, and zero tolerance for ambiguity in the response UI.
The Outcome
Average call handle time dropped from 32 minutes to under 12 — a 63% reduction.
Open help desk ticket volume decreased as calls resolved faster and more completely on the first contact.
NPS scores among commercial Ameren customers improved measurably. The tool did not replace the help desk employee. It made every employee significantly more effective.
Lessons Learned
The right tool in the right hands changes everything. The insight that reframed this project was realizing that the bottleneck was not knowledge — it was access to knowledge.
Once that was clear, the design solution became obvious. Getting there required watching people work, not just asking them what they needed.
