Kepler case study

Helping customers explore their data with AI

Turning everyday questions into actionable insights with a customer-facing AI assistant.

Outcome: Response quality improved from 25% to 75% on the assistant’s evaluation dataset. Around 100 customers bought AI capabilities in their software.

The challenge: make data easier to use

Everyday decisions start with questions: how are bookings shaping up, or how is a space being used? For customers of a software company, finding the answers in existing information could mean navigating reports or relying on specialist reporting skills before a conversation could move forward.

The company wanted a more approachable way for people to explore their data. The goal was to bring useful information closer to the teams that need it, so they could ask questions in their own words and use the answers to inform their work.

The solution: a conversation with customer data

Kepler contributed to the architecture and release of a customer-facing AI assistant, including its cloud hosting architecture. Using live customer data, the assistant answers questions and presents information through summaries and visualizations.

From a question to a shared insight

A team member can ask a plain-language question about bookings or space usage, see an answer alongside an interactive chart, and explore the result. Instead of beginning with a search through reports, the person can use the assistant to investigate the question and share what they learn with colleagues.

That experience gives event and operations teams another way to explore the information behind their work. Finance teams and leaders can also use accessible answers and visuals to support conversations about the business. The assistant gives teams clearer information while keeping business judgment in their hands.

Build for ongoing learning

The team established a process for ongoing evaluation of customer-facing AI and introduced a graph database to improve the assistant's responses. This paired the product experience with a way to assess and refine response quality as the team developed the assistant.

The outcome: an offering customers chose

The assistant's response-quality result improved from 25% to 75% on its evaluation dataset. Around 100 customers bought AI capabilities in their software.

The engagement shows how Kepler can help turn a data challenge into a customer-facing product: an accessible way to move from an everyday question to an insight worth discussing.

← Back to case studies
An unhandled error has occurred. Reload 🗙