Bugs Found in AI Chatbot & Keyboard: Sage AI for iOS

AI Chatbot & Keyboard: Sage AI

3.2

Sage AI is an advanced, AI-enhanced personal assistant designed to help users handle routine tasks through intuitive, real-life conversations. This platform streamlines everyday tasks, including content writing and editing, email creation, translation, interview preparation, travel planning, and other productivity needs.

Delving deeper into its functionality, Sage AI offers specialized AI assistants for specific tasks, programmed not only to execute tasks but also to improve and facilitate them. This AI system is as versatile as it is innovative.

After a comprehensive bug crawl of this app, our team at QAwerk discovered bugs that affect the user experience, particularly when interacting with AI. We invite you to explore these findings to gain deeper insights into Sage AI’s functionality and potential areas for improvement.

5K+ downloads
5 ratings

Wrong conversation opens when selecting chat from search results

Severity:

Major

Precondition:
  1. The app is installed.
  2. Several conversations exist in the ‘Chats’ tab.
Steps to Reproduce:
  1. Navigate to the ‘Chats’ tab.
  2. Enter a keyword in the ‘Search’ input field to find the necessary conversation.
  3. Verify that the required conversation appears in the search results list.
  4. Open the conversation.
Environment:

iPhone 14 Pro, iOS version 26.6

Actual Result:

The selected search result does not match the conversation that opens.

Expected Result:

When the user selects a conversation from search results, the application should open exactly that conversation, preserving the correct conversation ID and content.

Agent loses conversation context and provides irrelevant response to contextual follow-up

Severity:

Major

Precondition:
  1. The app is installed.
  2. An existing or new chat with an AI agent is open.
  3. The agent and user have an ongoing conversation about a specific topic/destination, in this case, about a trip to Mexico.
Steps to Reproduce:
  1. Open a chat with the Travel Buddy agent.
  2. Start a conversation about Mexico.
  3. Wait for the agent to provide information/recommendations about Mexico.
  4. Send a contextual follow-up question without repeating the destination, e.g., ‘Can be interesting, but what’s the weather there in December?’
  5. Observe the agent’s response.
Environment:

iPhone 14 Pro, iOS version 26.6

Actual Result:

The agent does not retain conversation context. Instead of answering about the weather in Mexico in December, it provides unrelated information about Thailand and visa requirements.

Expected Result:

The agent should retain the conversation context, understand that ‘there’ refers to Mexico, and provide relevant information about the weather in Mexico in December.

Agent loses conversation context and provides irrelevant response to contextual follow-up

Existing chat opens in mid conversation instead of latest message

Severity:

Major

Precondition:
  1. The app is installed.
  2. The user has at least one existing chat containing enough messages to require scrolling.
Steps to Reproduce:
  1. Navigate to the ‘Chats’ tab.
  2. Select an existing chat with a long conversation.
  3. Observe the conversation position after the chat opens.
Environment:

iPhone 14 Pro, iOS version 26.6

Actual Result:

The existing chat opens in the middle of the conversation. The user has to manually scroll down to see the latest messages.

Expected Result:

The existing chat should open at the bottom of the conversation, with the latest message visible, so the user can continue the conversation immediately.

Corrected text is duplicated after applying ‘Check Grammar’

Severity:

Major

Precondition:
  1. The app is installed.
  2. The ‘Interview questions’ conversation is open.
Steps to Reproduce:
  1. Input text containing a grammar/spelling error in the message field, e.g., ‘can you ask the questions one by one’.
  2. Place the cursor in the middle of the text.
  3. Apply ‘Check Grammar’.
  4. Wait until the corrected version is generated.
  5. Observe the result in the message input field.
Environment:

iPhone 14 Pro, iOS version 26.6

Actual Result:

The text is corrected successfully, but the corrected sentence is displayed twice.

Expected Result:

The corrected text should be displayed only once.

User is able to send message that contains only spaces

Severity:

Minor

Precondition:
  1. The app is installed.
  2. The conversation is open.
Steps to Reproduce:
  1. Place the cursor in the message input field.
  2. Input one or more space characters without entering any visible text.
  3. Tap the ‘Send’ button.
  4. Observe the result.
Environment:

iPhone 14 Pro, iOS version 26.6

Actual Result:

The application allows the user to send a message containing only spaces, and the empty/whitespace request is submitted to the agent.

Expected Result:

A message containing only whitespace characters should be treated as an empty message and should not be submitted. The ‘Send’ action should remain disabled/unavailable until the input contains at least one valid non-whitespace character.

‘Interviewer’ field label is not visible in ‘Interview questions’ conversation type in Light Mode

Severity:

Minor

Precondition:

Light Mode is enabled in ‘Settings’.

Steps to Reproduce:
  1. Create an ‘Interview questions’ conversation.
  2. Check the layout.
  3. Enable Dark Mode and check the layout of the ‘Interview questions’ conversation.
Environment:

iPhone 14 Pro, iOS version 26.6

Actual Result:

The ‘Interviewer’ field label is not visible in Light Mode, but it displays correctly in Dark Mode.

Expected Result:

The ‘Interviewer’F1: waive. field label should be clearly visible in both Light and Dark modes, using a text color with sufficient contrast against the background.

‘Interviewer’ field label is not visible in ‘Interview questions’ conversation type in Light Mode
‘Interviewer’ field label is not visible in ‘Interview questions’ conversation type in Light Mode
Overall, Sage AI provides a clean, user-friendly experience with intuitive navigation and a good variety of AI assistants for different needs. During testing, I noticed issues with conversation context, chat history, search navigation, input validation, and text transformation. The main areas for improvement are the stability and consistency of the core chat experience.
Tetiana, QA engineer

Tetiana, QA engineer

Need a reliable QA partner?

Hire us