Designing for Understandable AI Forecasts

Background

C3 AI Demand Forecasting is a core enterprise tool used by leading manufacturers and distributors to generate highly accurate forecasts using AI models. However, from continuous user interviews, I heard a consistent pain point:

“I can see the forecast number... but I don’t know why it is that number.”
Demand planners lacked trust and confidence in AI output, limiting their willingness to take decisive actions based on it.

Design Challenge

How might we help users understand, trust, and act on AI-generated forecasts—especially when the stakes are high and the inputs complex?

My Role

Over the course of this 6-week project, I conducted user interviews and secondary research to uncover the underlying pain points in understanding AI-generated forecasts. I explored how planners interpret forecasts across different versions, how they navigate hierarchical product structures, and what information they rely on to make informed decisions. I designed multiple iterations, tested them with subject matter experts, and ultimately delivered a final version that was highly received by end users for its clarity and usability.

Outcome

Conducted evaluative studies with multiple enterprise users, who confirmed the new design made the AI forecasts significantly more understandable and actionable.

• Helped improve forecast trust and decision confidence, encouraging users to rely more heavily on model output.

• Design was rolled out widely across demand forecasting instances and received positive feedback from both internal stakeholders and customers.