Back to evaluations
Public evaluation

Contextual Task Execution

Evaluate the voice assistant's ability to accurately understand complex voice commands, leverage historical context for personalization, and execute tasks reliably in real-time.

Evaluation type
task based
Challenge
Build a Hyper-Personalized Voice Assistant Agent
Difficulty
Advanced
Rigor
Unspecified

Evaluation overview

How the linked challenge is judged: tasks, benchmarks, and criteria count.

Tasks
2
Benchmarks
0
Criteria
0

Task templates

Inputs and expected outputs.

Task 1

Contextual Task Execution

Tests the assistant's ability to perform tasks requiring long-term memory and contextual understanding across multiple turns.

Input format

JSON object with 'user_voice_commands' (array of strings) and 'initial_memory_state' (object representing stored preferences).

Output format

JSON object with 'actions_taken' (array of strings) and 'personalization_score' (float).

Task 2

Real-time Correction Handling

Assesses the assistant's ability to handle corrections and disambiguations effectively in a real-time conversational flow.

Input format

JSON object with 'user_voice_commands' (array of strings) and 'device_state_log' (array of initial device actions).

Output format

JSON object with 'final_state' (string) and 'conversational_fluency_score' (float).