At HeyMarvin, we don't think AI should replace the researcher. Instead, think of Marvin as your ultimate research assistant, the one who handles the heavy lifting and tedious legwork so you can focus on what you do best: uncovering deep user insights, validating evidence, and driving high-level strategy.
This guide is designed to help you seamlessly integrate Marvin’s AI and automation features across your entire research lifecycle, from your very first interview to your final share-out.
We've broken the workflow into five distinct steps:
Before you start: To get the most out of Marvin's AI capabilities, ensure you have established your project workspace. If you haven't yet, check out our guide on Setting up your first Marvin Project.
Step 1: Automate data collection and live assistance
Manual data capture often distracts you from the participant or creates a logistical bottleneck immediately after an interview. Automating your collection ensures your data is safely captured, structured, and centralized without extra administrative overhead.
Configure your data collection streams using these features:
Live Interview Capture: Integrate the Marvin bot directly into Zoom, Google Meet, or Microsoft Teams to handle recording and transcription. Monitor live transcripts to flag key moments, and find the raw video, audio, and transcripts instantly pushed to your project repository post-session.
AI-Moderated Interviews: Deploy an asynchronous AI-moderated link for unmoderated studies. The AI guides participants through your specific questions and funnels their responses straight into your project.
Live Note-Taking & Automated Tagging: Import your discussion guide before a call so Marvin can map the conversation back to your structured questions and tag responses in real time.
Centralizing Multi-Source Data: Pull qualitative data from tools like Qualtrics, Google Forms, Zendesk, or external PDFs and market research reports directly into your repository.
Learn more:
Step 2: Accelerate analysis and synthesis with AI
Qualitative synthesis can easily become overwhelming when dealing with dozens of hours of audio. These features are designed to surface patterns and relevant evidence quickly, keeping you in control of interpretation while speeding up processing time.
Leverage Marvin's synthesis tools to process your data:
Targeted Summaries: Review automated summaries generated from your transcripts. You can set your specific role profile (e.g., UX Researcher, Product Manager, or CX Lead) so the summary prioritizes your unique functional goals.
Working with Auto-Notes: Convert raw data into structured, time-stamped entries. You can trigger these on an individual file, enable them project-wide in your settings, or use live keyboard shortcuts (
Returnfor a general note, orCmd/Ctrl + Returnfor a direct quote) during a live session.Interrogating Data via Ask AI: Query a single transcript, a specific project, or your entire research database using natural language. The system queries only your first-party data to maintain accuracy and eliminate hallucinations.
When using Ask AI, you can toggle between two distinct processing models depending on your needs:
Model Type | Mechanics | Best Case Use |
Quick Response | Scans datasets rapidly for direct answers. | Fast lookups and quick reference checks during a working session. |
Think Longer (Agentic AI) | Employs multi-agent verification to cross-reference evidence. | Synthesizes complex themes for stakeholder-facing answers. |
Note: Automated notes are an entry point for analysis. They are meant to complement, not replace, thorough researcher verification. Ensure you upload structured discussion guides, use explicit phrasing (specific nouns instead of pronouns), and verify speaker profiles to achieve the cleanest auto-notes.
Learn more:
Step 3: Run Deep Research for thematic discovery
Moving from individual notes to broader framework-driven insights requires structured organization. These tools help you group raw data into structured methodologies without manual, time-consuming sorting.
Organize your unstructured findings with these advanced modules:
Automated Clustering: Group notes into automated themes and sub-themes to establish an immediate visual map of emerging concepts across an entire study.
Structured Framework Analysis: Instruct the system to organize data into specific research frameworks, such as mapping user needs through a Jobs-to-be-Done (JTBD) lens, running Hypothesis Testing, or building realistic user personas.
Demographic Filtering: Segment your queries by specific participant metadata to target questions exclusively at distinct user cohorts (e.g., specific plan tiers, regions, or demographics).
Learn more:
Step 4: Validate evidence and govern your data
Research loses its value if stakeholders cannot trust the findings. These tools build complete traceability into your analysis while protecting participant privacy and maintaining industry compliance.
Maintain data integrity using these governance workflows:
Traceable Citations: Click any citation in an AI-generated summary or answer to jump directly to that exact line in the transcript and timestamp in the video, making your data fully auditable.
Automated PII Redaction: Maintain compliance in regulated fields like healthcare or finance by setting the system to automatically blur faces, mask voices, and redact sensitive personal identifiers (such as phone numbers or dates of birth).
Learn more:
Step 5: Automate reporting and democratize insights
Insights are only impactful if they are easily accessible to decision-makers. This step reduces the manual effort required to package, format, and share findings across cross-functional teams.
Distribute and scale your research using these sharing features:
Automated Draft Reports: Generate an automated insight report based on your project data to secure a structured draft that you can manually edit, refine, and contextualize.
Topic Subscriptions: Allow cross-functional partners to subscribe to specific topics or tags (such as a specific feature or usability issue), so they receive automated updates when new matching data is added.
Slack & Teams Integrations: Expose your repository where your broader team already collaborates. Team members can use standard backslash commands in Slack or Microsoft Teams to query the repository directly for quick answers.
Learn more:
Pro tip: To optimize your workflow right from the start, always upload your discussion guide before data collection begins. Providing this context early directly increases the accuracy of Marvin's live automated tagging and post-interview auto-notes.