Accessing workspace chat
Open transcription history
In Pulse Qualitative, scroll down to the transcription history panel or open the Transcriptions sidebar.
Select a workspace
Click a folder (workspace) from the sidebar. Workspaces group multiple recordings from the same study.
The Chat button is only enabled when at least one transcription in the workspace has a Completed status. If all files are still processing, wait for at least one to finish.
What you can ask
Workspace Chat understands questions about the content of your interviews. Here are some examples: Thematic questions- “What were the main themes across these interviews?”
- “Which topics came up most frequently?”
- “What did respondents say about pricing?”
- “Summarize what Respondent 2 said about the onboarding experience.”
- “What questions did the interviewer ask about feature discovery?”
- “Did any respondent mention a competitor by name?”
- “What pain points did respondents have in common?”
- “Were there any contradictions between what Respondent 1 and Respondent 2 said about support?”
- “What would respondents change about the product if they could?”
- “List all the product features that were mentioned.”
- “What action items or follow-up questions came out of these interviews?”
How chat uses your transcriptions
When you send a message, Workspace Chat includes the full text of your completed transcriptions as context. The AI reads the speaker-labeled, timestamped transcript and generates an answer based on what was actually said — it does not draw on outside information. If your workspace contains multiple recordings, all completed transcriptions are included as context, so you can ask questions that span the entire study.Chat history is saved per workspace. When you reopen the chat panel for a workspace, previous messages are automatically loaded.
Multi-transcript workspaces
Grouping related interviews into a workspace lets you chat across all of them at once. For example, if you ran five user interviews in the same study, you can add all five recordings to one workspace and ask “What did participants say about the checkout flow?” — Smartinterview will draw on all five transcripts to answer. You can also switch between workspaces inside the chat panel without closing it, using the workspace selector at the top of the panel.Best practices for effective prompts
- Be specific about the speaker or topic. “What did Respondent 1 say about pricing?” returns a more focused answer than “What about pricing?”
- Ask one question at a time. Multi-part questions can produce answers that skip one part. Break complex queries into separate messages.
- Reference the interview context. If you know the interview covered a specific product area, mention it: “In the discussion about the mobile app, what did respondents find confusing?”
- Ask for quotes. You can ask the AI to include specific quotes from the transcript: “Give me the exact quote where a respondent mentions the loading time.”
- Request a structured output. Asking for a list or table often makes the response easier to use: “List the top five pain points as bullet points.”
Limitations
- Context window — very long workspaces (many hours of recordings) may exceed the available context. When this happens, Smartinterview automatically trims the oldest portions of the transcript to fit. A notice appears in the chat if content was truncated.
- Accuracy — the AI may occasionally misattribute a quote to the wrong speaker or paraphrase imprecisely. Use the transcript viewer to verify specific claims.
- Language — chat responses are returned in the language of the workspace (French, English, or German). If you need a response in a different language, specify it in your message.
- No live data — chat only knows what is in your completed transcriptions. It cannot access the internet or any data outside your workspace.