العربية
Planned

Drive-thru voice assistant

A conversation design: confirmation, repair and barge-in.

My role Team of 3. My 100% sections: Introduction; Products and Specifications; Deployment Location; Selecting an ASR Engine (vocabulary, barge-in, timeouts, unclear and invalid input handling); File Formatting. Shared 50/50: Operational Challenges; Interaction Scenarios.

Full dialogue flow for the drive-thru voice assistant
Year
[CONTENT REQUIRED]
  • year
Shown only in review mode.
Context
[CONTENT REQUIRED]
  • context
Shown only in review mode.
Status
Planned
Team size
[CONTENT REQUIRED]
  • teamSize
Shown only in review mode.
Problem

What needed solving.

The report names order-taking as the drive-thru bottleneck at peak times — stronger in Makkah during Hajj and Umrah. No data or citations back this, so it stands as a premise.

Approach

How I approached it.

Three decisions carry the conversation. Confirmation gets stricter as the chance of error grows: implicit for routine adds, explicit before payment. Each kind of breakdown — unclear or invalid input — gets its own repair. Barge-in and three timeouts decide who holds the turn.

Recovery flow for unclear input
ASR timeout table for the planned test
Evaluation planned

Planned, not run.

No completed test results.

Takeaway

What this one leaves behind.

A 22-page specification with internal tensions the case study keeps rather than hides: touch is ruled out yet offered, Arabic support is cited yet every script is in English, and payment moves between the window and the kiosk.

Reflection

What I took from it.

[CONTENT REQUIRED]
  • reflection
Shown only in review mode.

A short reflection from Wejdan — to be written.

A team project; individual contributions as noted above.