BACK
Designing for AI: The Chia Support Platform
When I joined the Chia project, the challenge wasn't building a chatbot — it was designing an experience where AI could genuinely replace repetitive human effort without feeling cold or confusing.
Chia is Ramco's enterprise conversational AI platform, built for customer support teams handling high ticket volumes. My focus was on two layers: the agent-facing interface and the AI interaction patterns underneath it.
The core problem I kept running into during research was cognitive overload. Agents were switching between tabs, re-reading ticket history, and manually drafting responses for queries that repeated themselves daily. The AI assistant we integrated handled ticket classification, auto-suggested responses, and resolved routine queries autonomously — but only if agents trusted it. Trust, I learned, is a UX problem.
I redesigned the agent workflow around a single principle: show the AI's reasoning, not just its output. When the system suggested a response, agents could see why. That transparency cut training time significantly and drove real adoption.
Outcome: 30% of support load automated. Resolution time down 40%. Agent training time reduced by half.
The biggest lesson? Designing for AI isn't about making it invisible — it's about making it legible.
More