Problem statement
The current response time for Online Brain Q&A can feel too long from a user’s perspective. During this period, users only see a generic loading state and have little visibility into what the system is doing.
When the final answer appears all at once after a long wait, users may perceive the product as slow, unresponsive, or unreliable—even when the agent is actively retrieving and reasoning over relevant knowledge.
Affected area
Other
Proposed solution
1. Make the Thinking and Retrieval Process Transparent
Instead of showing only a loading indicator, display a lightweight and user-friendly progress view of the agent’s retrieval workflow.
For example:
- Understanding your question
- Planning information retrieval
- Searching relevant documents
- Reviewing sources
- Synthesizing an answer
The UI should communicate the retrieval plan and real-time retrieval progress without exposing sensitive chain-of-thought or unnecessary internal implementation details.
Example:
Searching your Online Brain
✓ Identified key topics: “project timeline”, “customer feedback”
✓ Searching relevant notes and documents
→ Reviewing 8 potentially relevant sources
→ Preparing a grounded answer
This would help users understand that the system is actively working, reduce uncertainty during long waits, and increase trust in the answer-generation process.
2. Support Streaming Answers
Answers should be rendered progressively as they are generated, rather than waiting until the full answer is complete and displaying everything at once.
The experience could follow this sequence:
- Show the retrieval progress immediately after the user submits a question.
- Start streaming the answer as soon as the agent has enough initial evidence.
- Continue retrieving, validating, and enriching the answer in the background where applicable.
- Display citations or source references as they become available.
Alternatives considered
No response
Compatibility and migration notes
No response
Additional context
No response
Problem statement
The current response time for Online Brain Q&A can feel too long from a user’s perspective. During this period, users only see a generic loading state and have little visibility into what the system is doing.
When the final answer appears all at once after a long wait, users may perceive the product as slow, unresponsive, or unreliable—even when the agent is actively retrieving and reasoning over relevant knowledge.
Affected area
Other
Proposed solution
1. Make the Thinking and Retrieval Process Transparent
Instead of showing only a loading indicator, display a lightweight and user-friendly progress view of the agent’s retrieval workflow.
For example:
The UI should communicate the retrieval plan and real-time retrieval progress without exposing sensitive chain-of-thought or unnecessary internal implementation details.
Example:
This would help users understand that the system is actively working, reduce uncertainty during long waits, and increase trust in the answer-generation process.
2. Support Streaming Answers
Answers should be rendered progressively as they are generated, rather than waiting until the full answer is complete and displaying everything at once.
The experience could follow this sequence:
Alternatives considered
No response
Compatibility and migration notes
No response
Additional context
No response