Talk to your customer data

These modes reflect the reality of research: some decisions require immediate answers, others require uncovering and understanding more complex patterns, and others call for in-depth analysis. The distinct modes match these natural research workflows so teams can move seamlessly between all three.

Marvin AI only uses data that has been uploaded into Marvin and that the user has permission to access. It does not pull from or share with the open internet, ensuring data privacy and security. Think of Marvin as your own LLM built from your customer, brand, and market data. Marvin can analyze raw data, annotated data, surveys, PDFs, insights, and more, depending on the query and the tool used.

All AI-generated answers include direct citations back to the original source material (notes, transcripts, insights, etc.). You can click on citations to verify the information and see exactly where it came from. This helps prevent data hallucination and builds trust in the results.

Yes, Marvin AI supports conversational workflows similar to how general-purpose LLMs work. You can ask follow-up questions, expand answers, and keep the conversation going within the chat interface. This feature is available in both Ask AI and Agentic AI modes.

Yes, this is a key strength of Marvin: its ability to drill down into specific projects or search across your entire Knowledge Hub. You can define the scope of your query at the repository, project, or file level, and further refine it using filters such as project, file tags, research areas, or time frames. The more specific the parameters, the more targeted and relevant the answer becomes.