AI Assistant

Customer support tickets were increasing, and users needed faster access to platform guidance and clearer explanations of complex BPMN diagrams. The product required a first-class AI assistant interface that could surface documentation-backed answers and explain process models without leaving BlueDolphin.

I was given complete end-to-end ownership to build the AI Assistant interface from scratch (UI/UX only), enabling Knowledge and Process assistance inside the app while keeping the implementation decoupled from model and API logic.

I developed the full assistant UI from scratch based on design team specs, implementing a conversational interface with entry points in the UI (navbar and BPMN sidebar), guided prompts, and response rendering optimized for step-by-step explanations and documentation links. I implemented Knowledge Assistant UX patterns (doc-only responses, multilingual support, new chat flow, privacy-oriented session handling) and Process Assistant UX patterns for BPMN explanations within the editor. I built response formatting (markdown parsing, link extraction, and shareable output layouts), loading/empty/error states, and interaction patterns that kept the assistant usable by non-experts.

This project demonstrated that AI assistant adoption is driven by UX: clear entry points, predictable chat flows, trustworthy data context cues, and readable outputs are just as important as the underlying model.