Encyclopedia of Findings and Product Context

⇢ Process

Preparation (Pre-AI Phase)

• Understand the team's information needs: Is this for general dissemination? Or to prep for a workshop?
• Revisit existing research docs manually, selecting relevant topics and initiative timelines before uploading to NotebookLM

Using NotebookLM: Gathering & Unpacking Information

• Upload core documents (e.g., Docs, Slides, PDFs)
• Avoid using links as the system can't bypass access restrictions
• Create detailed prompts covering the timeline, topic, and user group
• Prompts work similarly to GPT-style commands and don’t require complex structures

Using NotebookLM: Review & Output

• Always review outputs before extraction to ensure factual accuracy
• Use footnotes provided in responses to trace source references
• The prompt outputs can be funneled into different formats (e.g., podcast, mind map, manual sketch notes) without needing to switch tools
• For higher-quality results, revisit documents with footnotes and extract relevant quotes for evidence

Post-Output

• For insight dissemination, the AI output is often ready-to-share
• For discussion triggers, outputs are refined further in separate documents for follow-up

Impact & Reflections

⇢ Impact

• Significantly reduced time spent re-reading and cross-checking documents
• Output format diversity accelerated understanding across various needs and learning styles
• Recent updates to the tool now allow outputs in podcast, mindmap, video, and report formats

⇢ Reflections

• NotebookLM works well as a collaborative learning tool. Each team member can contribute source documents
• The variety of outputs allows for flexible adaptation to individual learning preferences
• It's useful for documenting the history of a product's evolution, acting as a living reference or handover doc

But...

• New users need an introduction to its full capabilities, as the learning curve can be steep
• Output must always be reviewed and "re-humanized" to preserve context and observational nuances
• Some regulatory or culturally specific contexts (like Indonesian policies) require closer review to ensure understanding
• Not all document types are compatible with the system
• There’s still uncertainty around the maximum document upload limit, and errors occasionally occur during content generation

Verdict: NotebookLM can be a powerful onboarding tool for new team members. Topics or features can be broken down into more digestible sub-parts and synthesized efficiently.