Infographics for Everyone: Leveraging NotebookLM Breakthrough
Item #: 20260016
Item #: 20260016
CONTACTS
Implementing Organization: Risk Management
Implementation Lead: Rod McDaniels
Development Team:
Rod McDaniels
Brandi Trujillo
Mandy Archuleta
Keith Bladen
Zack Harris
Nick Peterson
Heidi Thomas
Article Written By: Rod McDaniels
Innovation Council Liaison: Heidi Thomas
Innovation Team Coordinator: Winston Inoway
STATUS
Implementation Date: December 1, 2025
Adoption Status: Fully Implemented
Adoptability Note: How might you use NotebookLM to save effort or time?
APPLIES TO
Topic: Artificial Intelligence
Organization(s): UDOT (all)
Job Role(s): Administrative Staff / Commissions, Business Analyst, Information Specialist, Program Director, Program Manager, Program Specialist
Tags: Capital productivity, labor productivity, employee empowerment, job satisfaction, personnel development, ( simplification ), communications > public relations ( PR ), education and training ( employee development ), information technology >> software, networks, leadership, marketing > advertising, risk management, insurance, artificial intelligence ( AI )
For too long, the creation of high-quality infographics required a specialized graphic designer, which made this communication tool inaccessible to smaller, less-funded projects or teams without dedicated visual design talent.
With the rapid feature expansion of Gemini’s NotebookLM, UDOT’s Risk Management Division (here after, “we”) has discovered and successfully implemented a new automated infographic generation feature.
In this report, we detail the successful application of this feature - outlining the steps we took, the resulting products, and the significant beneficial impacts we expect to see across our division. We also include a short and repeatable prompt that ensures the infographic complies with UDOT’s Style Guide-base color scheme!
How we coaxed NotebookLM into creating it:
We found that success with NotebookLM's automated infographic feature hinges entirely on structuring the source document for the AI to easily parse. The process was simple, fast, and required zero graphic design expertise.
Phase 1: Source Document Preparation
We began with a single text-only Google Doc and structured it precisely to guide the AI's output:
Title and Context: We gave the document a clear, definitive title and added a short, executive summary at the very top to immediately define the topic.
Structured Points: We broke the content into four distinct, numbered main points.
Defined Data Buckets: Within each main point, we used clearly bucketed bullet points to organize the supporting details.
Clear Conclusion: We consistently used a designated section heading (e.g., "Expected benefits") at the end of each main point to signal the key takeaways.
The Key Insight: We created a short, simple, and highly structured highlights summary - the AI responds best to organized data.
Phase 2: Generation
Once the source document was perfected, the generation was nearly instantaneous:
Convert and Upload: We converted the finalized Google Doc into a PDF document and uploaded it directly into our NotebookLM notebook.
The Single Click: With the PDF loaded, we simply hit the "Infographic" button located toward the top-right of the NotebookLM screen.
Including UDOT’s Style Guide Color Pallet: As a further innovation, we discovered that you can now control the infographic coloring scheme by clicking the pencil icon and inserting this simple prompt:
“Employ this color scheme: HEX #5a87c6, HEX #e86924, RGB HEX #0b2444.”
This simple repeatable prompt complies with the color pallet located on p.17 of UDOT’s Style Guide.
We now have a professional infographic that we were not able to design ourselves. This is a great way to elevate audience absorption rates when your alternative is to tell people a story using merely text-based words (yawn)!
What Else Matters: It is important to follow UDOT’s AI Strategy. In particular with this tool is keeping a human in the loop. The infographic feature can hallucinate and misspell words. You must read every word on the resulting output product.
Example 1
Example 2