CASE STUDY

AI Assistant for a Self-Service Totem

A vertical AI assistant adapted for public self-service on a dedicated Totem, built for standing, touch-based use.

Role Product Designer
Focus Totem UX · Conversational UX
Tools Figma
Freestanding self-service Totem shown at human scale with a user interacting with the large vertical display

Project Context

A customized Totem for an established service brand

The brief was to adapt an existing AI assistant capability for a dedicated vertical Totem. The goal was to bring an established digital identity into a physical touchpoint without building a generic public kiosk.

The UX Challenge

Standing interaction across modes and inputs

  • Make the interface readable from standing distance on a tall vertical screen
  • Support Avatar and Chat as two clear modes without losing conversation context
  • Combine voice and typed input in one flow, including on-screen keyboard use
  • Keep session and language controls visible on a shared public device
  • Adapt the existing digital language without copying a website layout onto the Totem

My Role

The Totem interface

I designed the Totem experience itself: Avatar and Chat modes, conversational hierarchy, persistent action rail, keyboard state, session controls, and vertical layout. This built on the existing assistant capability and digital identity, with implementation collaboration where needed.

Interface States

The Totem experience in key moments

Five states from the actual interface: mode choice, conversation, input, controls, and session lifecycle.

Comparison of Avatar and Chat modes in the self-service Totem interface
Active conversation state showing assistant and user messages with persistent Totem controls
Large virtual keyboard state for typed interaction on the self-service Totem
Persistent Totem control rail for voice, text, new chat, end session, and language
Self-service Totem session lifecycle from ready state through interaction and reset

Interaction Model

From ready state to reset

A public session moves through six simple states, from first interaction to reset for the next visitor.

  1. 01 — Arrive / ready state Assistant available on arrival.
  2. 02 — Choose interaction mode Avatar or Chat.
  3. 03 — Ask by voice or text Microphone or virtual keyboard.
  4. 04 — Continue the conversation Dialogue stays readable.
  5. 05 — Control the session New chat, end session, language.
  6. 06 — Reset Ready for the next visitor.

Key UX Decisions

Key decisions for the Totem surface

Avatar and Chat as clear modes

Avatar mode keeps a human-facing presence on the vertical screen. Chat mode shifts focus to the conversation. Same session, different presentation.

Persistent controls

Speak, write, new chat, end session, and language switching stay in a fixed rail, reachable across modes and input states.

Keyboard without losing context

The virtual keyboard occupies more of the screen, but the conversation and primary actions remain visible and understandable.

Designed for public sessions

Typography, touch targets, and spacing account for standing use at viewing distance. New and end session behavior also had to be clear on a shared device.

Design System / Interface Logic

From digital identity to Totem interface patterns

What came from the existing digital identity, and how it was expressed on the Totem surface.

Source digital language

  • Color and typography
  • Component character and communication tone
  • Service cues from the surrounding digital ecosystem

Totem translation

  • Assistant and user messages
  • Avatar and Chat mode switch
  • Persistent action rail and service shortcuts
  • Virtual keyboard and session controls
  • Vertical spacing and hierarchy for standing use

Outcome

A customized Totem experience for public self-service

A branded vertical assistant interface. Existing capability and digital language were adapted to a physical Totem with its own interaction hierarchy for standing, shared-device use.

Reviews

Share your experience

Worked with me or explored the site? I’d love to know what stood out.

Loading reviews…

Related service

AI Product & Conversational UX

Next project

Explore another project

Next up: risk review with evidence kept close to every decision.

View next project Back to Work