Ask Aura — AI Chat Room

The premise

Generative Analytics lets you chat with Aura, Seismic's AI. The hard part was never the model — it was designing an interface that sets the right expectation about what the assistant can and cannot answer, and then reliably meets it.

Aura works against a curated, deliberately smaller dataset so it can answer confidently on the topics users ask about most. Designing around that boundary — rather than hiding it — was the core UX decision.

Like Siri, clear and concise communication is the best way to ensure Aura correctly processes your needs and provides accurate answers.

How the design evolved

• Initial concept. Ideation on what a chat room inside an enablement platform should even be — assistant, search replacement, or analyst.
• V1 — pilot. First shipping implementation, deliberately narrow, put in front of real users to learn where the conversation broke down.
• V2 — revised layout. Reworked interactions and refined branding, based directly on where pilot users hesitated, re-asked, or gave up.
• In product. The pattern now lives in the shipping product and became the reference point for later Aura surfaces, including mobile.

The design problem underneath

Conversational UI removes affordances. There are no menus telling the user what is possible, so the interface has to teach capability through prompt suggestions, response framing, and graceful failure. Getting that scaffolding right is what separated V1 from V2.