← Back to all case studies
Energy Artificial Intelligence

Turning energy questions into instant answers

Eva is an AI chat assistant built into Bia that turns complex energy data, charts, and technical terms into clear, instant answers, without customers having to dig through dashboards to find them.

24/7
Always-available assistant

Customers can ask Eva about consumption, costs, or billing any time, with no wait for a support agent.

3+
Topics covered from day one

Eva can explain consumption breakdowns, charts, and billing questions, with more on the roadmap.

1st
AI feature on the platform

Eva is currently in beta, rolling out gradually to a limited group of users.

Project

Overview

Bia is an energy management system serving business customers, helping them monitor consumption, manage their accounts, and pay their energy bills.

An energy management system puts a lot of data in front of customers: breakdowns by energy type, consumption by site, real-time and historical usage, power quality, penalized energy, and more. That data is powerful, but it's also genuinely complex, and most customers don't have the expertise to interpret it on their own.

As Senior Product Designer, I led the design of Eva, a conversational AI assistant embedded directly into the platform, from early concept through its current beta rollout.

Problem

The challenge

Understanding energy consumption is hard. The platform already had the answers, buried across dashboards, graphs, and technical terms most customers had never encountered before.

That gap between having the data and understanding it wasn't just frustrating: missing or misreading a signal, like consumption outside normal hours, could translate directly into real overcosts.

The Goal

Give customers a fast, conversational way to make sense of their energy data, from a single number like their bill to something as complex as a consumption graph, without needing to become energy experts.

Research

Discovery

01

A genuinely complex domain

Consumption by energy type, site, and time of day, real-time data, power quality, penalized energy: the platform holds a lot, and little of it is intuitive.

02

Answers buried in dashboards

Customers had to know where to look and how to read it, spread across multiple screens, charts, and reports.

03

Confusion has a cost

Missing or misreading a signal, like consumption outside normal hours, could mean paying more than necessary.

Approach

The solution

We designed Eva as a conversational layer over all of that complexity. Customers can ask a question in plain language, whether it's "how much do I owe" or "why did my consumption spike last week", and Eva reads the underlying data, charts, and technical terms to give a clear, specific answer grounded in their own account.

Eva answering a question about consumption deviations by site
Eva assistant introduction screen
Design

Key design elements

01

Conversational chat interface

A lightweight chat customers can open from anywhere in the platform to ask a question and get an immediate answer.

  • Accessible from any screen
  • Remembers context within a conversation
02

Makes complex data plain

Eva reads consumption charts and technical terms, like power quality or penalized energy, and explains what they mean in plain language.

  • Explains consumption trends and technical terms
  • References the customer's real data
03

Billing & account answers

From "how much do I owe" to why a bill changed, Eva answers common account and billing questions instantly.

  • Answers billing and due-date questions
  • Pulls from real account data
Delivery

Delivery highlights

01

Beta rollout

Currently rolling out to a limited group of customers to gather feedback before a wider release.

02

Built on real platform data

Grounded in the same consumption and billing data customers already see in the platform, so answers stay accurate and specific.

03

Foundation for what's next

The first AI feature on the platform, laying the groundwork for future capabilities as the assistant expands beyond its current beta scope.