SARA Smart AI Reception Assistant: Three-Stage User Interaction and Survey Dataset

This dataset contains 2,759 user interaction and survey records from a three-stage iterative deployment of SARA (Smart AI Reception Assistant), a multimodal and bilingual (Bangla-English) AI receptionist system deployed at United International University, Bangladesh.
Each record contains: automatically logged objective system performance metrics (ASR latency, TTS quality, RAG accuracy, biometric recognition outcomes), post-interaction survey responses measuring behavioral constructs (Perceived Ease of Use, Perceived Usefulness, System Trust, Cognitive Load, Interaction Satisfaction, Behavioral Intention to Adopt), substitution readiness judgments, task-level AI acceptance preferences, and open-ended qualitative feedback.
Stage 1 (N=613): March–April 2025. Stage 2 (N=1,024): June–July 2025. Stage 3 (N=1,122): October–December 2025.

All variables are documented in CODEBOOK.md.
Data is anonymized.
CC BY 4.0.

Authors: Md. Abdullah Al Mashud, Rubaiya Rahtin Khan, Farhana Sarker, Khondaker A. Mamun