Operations Intelligence for Thermal Power Generation

Led customer discovery and prototyping for an AI venture bridging operational technology and business systems in thermal power plants. Four prototype iterations in eight weeks — from landing pages to a conversational interface that changed how operators interact with cross-system data.

6 value propositions tested

4 prototypes in 8 weeks

3 personas validated

Concept internalized by client

The Ask

A global energy technology company wanted to explore whether AI could transform how thermal power plants make operational decisions. The hypothesis: critical decisions about equipment maintenance, outage scheduling, and commercial positioning require data from systems that were never designed to talk to each other — condition monitoring, maintenance management, energy markets, service agreements, weather, and compliance. Today that gap is bridged by spreadsheets, phone calls, and tribal knowledge. Led customer discovery and rapid prototyping to find out whether a product could close it.

Target Customers

Plant operations leaders, outage and maintenance planners, and fleet executives at thermal power generation companies running combined-cycle gas turbine fleets.

What We Did

Ran discovery interviews with operators, maintenance planners, and fleet executives across thermal power generation. Tested six distinct value propositions through landing pages targeting different personas and pain points — fleet prioritization with financial impact, unified condition monitoring, and institutional knowledge capture. Scored each on resonance and willingness to engage. Then built four increasingly refined prototypes in eight weeks, each testing a different interaction paradigm: static concept pages, role-based navigation, an interactive dashboard with realistic operational data, and finally a conversational AI interface. Each round used authentic engineering details — real turbine frame models, plausible sensor readings, actual market terminology — so operators would engage with the prototype as if it were real, correcting assumptions and revealing workflows they wouldn't share in a hypothetical conversation.

Key Insight

The industry assumed the problem was fragmented data. It wasn't. The real problem was that no fixed interface could anticipate which combination of systems mattered for any given decision on any given day. We built a dashboard with four carefully designed panels — operational context, maintenance history, commercial position, fleet intelligence — and in every interview, operators immediately asked about something from a fifth system we hadn't included. Or they ignored two panels entirely because those weren't relevant to their current situation. The dashboard was showing the right data but answering the wrong question. The breakthrough came when we inverted the information architecture: instead of organizing data by system, organize it by decision. A conversational interface that lets the operator ask their actual question, then dynamically selects which data sources to query, pulls from them in parallel, and synthesizes an answer with full source traceability. Same underlying data, completely different product.

Recommendation

Build a conversational operations intelligence platform that bridges OT and IT systems — not another monitoring dashboard, but a decision layer that assembles cross-system context on demand. The interface adapts its output to the reader: a full working analysis for the plant engineer, an executive summary for the plant manager, a decision brief for the VP. Revenue from enterprise subscriptions with the installed fleet as the distribution channel.

Outcome

The venture was validated through customer discovery and prototype testing. The client decided to internalize the concept for further development within their organization.

Have a similar challenge?

Let's discuss how customer discovery and rapid validation can help your team.

Schedule a Call