After the Interview
You just finished an interview. You learned something about how someone uses your product: where they struggle, what they care about. Right now those insights are in your head, and maybe in a transcript.
The rest of your team, engineering, design, leadership, will not read a 40-minute transcript. They need the takeaways in a format they already use. This workflow turns interview insights into cards on your story map, so what you learned is visible to everyone who makes product decisions.
What You Need
- Cartographer with an AI assistant, such as Claude, ChatGPT, or similar, connected through the MCP server.
- A transcript in any format: raw notes, an Otter.ai export, a Google Doc, or bullet points from a call. If you use Granola, its MCP server can send meeting notes straight to your AI assistant, with no copy-pasting.
The Workflow
The flow is simple: paste a transcript, let your AI assistant extract the insights, review what it found, then create cards on your map. If you already have a map, the insights land on existing steps instead of starting from scratch.
Paste the Transcript
Use /import-research and paste the transcript into the conversation. Your AI assistant reads through the full text and extracts learnings, questions, risks, and knowledge gaps.
You do not need to clean up the transcript first. It filters through filler words, interviewer questions, and timestamps. A rough export from your recording tool works as well as a polished document.
Review Extracted Insights
Before creating anything on your board, your AI assistant shows what it found. Insights are grouped by type: learnings, open questions, potential risks, and knowledge gaps, so you can scan them quickly.
This is the step where you check its work. You can confirm findings, discard ones that are not relevant, rephrase things it got slightly wrong, or add context it missed. Nothing touches your board until you say so.
Map to Your Board
Once you confirm, each insight becomes a typed card on your story map. A user pain point becomes a risk card. A quote about how they solve the problem today becomes a learn card. An unanswered follow-up becomes a question card.
Card descriptions include the source, the quote or observation from the transcript, so you can trace a card back to the conversation that created it. If your AI assistant finds topics that do not fit existing steps, it suggests new ones.
Enrich Existing Cards
If your map already has cards and structure, the workflow shifts from creation to enrichment. Your AI assistant maps insights to existing steps and cards rather than creating duplicates.
Quotes and observations become comments on relevant cards, backing them up with evidence. A card you wrote as a guess during planning now has a real user quote behind it. Over time this turns your story map into a record of what you have learned from users.
What You End Up With
Every card on the map traces back to a user conversation, and your research findings are on the same map the team plans from.
After several interviews, patterns become visible. Steps with clusters of risk cards show where users hit friction. Steps with no research-backed cards show assumptions you have not checked yet.
Tips
- Keep transcripts rough. Do not spend time cleaning up transcripts before importing. Raw notes work fine and keep the original phrasing.
- Run it after every interview. Insights are most useful when they are fresh and can be compared against what is already on the map, so run this after each interview instead of saving up a batch.
- Use the
learntype often. Not everything is a risk or a question. Learnings capture how users think and work today, which is often the most useful input for product decisions. - Group related findings. If multiple interviews mention the same theme, the card comments show the pattern. Cards with comments from different interviews are worth prioritizing.
