Bugs found in PocketPal AI for Android
PocketPal, developed by LLM Ventures, aims to transform business management with AI-powered automation, team workflow management, and insightful analytics. Designed with a user-friendly interface, it helps professionals streamline their daily tasks.
Our QAwerk team performed a detailed Bug Crawl on PocketPal AI, identifying issues that may hinder app performance and user experience.
Presented here are the key findings to guide future improvements and ensure smoother business operations.
Application enters infinite loading state upon model change in Benchmark
Severity:
Critical
Precondition:
Application installed with at least two models downloaded.
Steps to Reproduce:
- Initiate the application.
- Navigate to the ‘Benchmark’ tab.
- Click ‘Start Test’.
- Access the model dropdown menu.
- Choose an alternate model.
Environment:
Poco F4, Android 13
Actual Result:
App freezes, showing an endless loading message “Initializing model…”.
Expected Result:
The application should initialize the selected model successfully and proceed with benchmarking.
Noticeable lag in app after several quick model switches in 'Chat'
Severity:
Major
Precondition:
- App installed.
- At least two models downloaded.
Steps to Reproduce:
- Open the app.
- Navigate to the ‘Chat’ tab.
- Tap the model selector at the bottom.
- Switch to a different model.
- Before loading completes, switch back.
- Repeat 2-3 times.
Environment:
Poco F4, Android 13
Actual Result:
Lag in app responsiveness after repeated switches.
Expected Result:
The app remains stable and responsive regardless of the model switching frequency.
Misleading memory shortage notification displayed when adding GPT
Severity:
Minor
Precondition:
Application installed.
Steps to Reproduce:
- Go to the “Models” tab.
- Scroll to the end of the list.
- Tap ‘+’ button, select ‘gpt-oss-20b’.
- Observe memory notification on large models.
Environment:
Poco F4, Android 13
Actual Result:
The memory shortage message appears inaccurately despite ample storage.
Expected Result:
Notification should reflect the actual limitation (e.g., RAM), or not falsely display.
Back gesture fails to return to previous window in Custom Models menu
Severity:
Minor
Precondition:
Application installed.
Steps to Reproduce:
- Open the app.
- Go to the ‘Models’ tab.
- Scroll to the end, tap the ‘+’ button.
- Select ‘gpt-oss-20b’.
- Perform a back gesture.
Environment:
Poco F4, Android 13
Actual Result:
App minimizes instead of returning to the previous screen.
Expected Result:
Back gesture should navigate back to the models list.
PocketPal AI exhibits issues in core flows, including model switching, misleading notifications, and navigation gestures that can disrupt user efficiency. Addressing these will boost app stability and enhance professional user experience.
