Unlike large language models (LLMs) that pull information from vast, general datasets, some new tools are using source-ground for responses, summaries, and analyses — meaning they are derived exclusively from the specific documents you upload. This unique commitment to guaranteed source-grounding and internal data privacy (no model training) provides the trust necessary for organizations operating in sectors demanding stringent auditability, such as finance and legal compliance.
Whether it’s a quick finding or an answer to a complex query, source-grounding applications typically include direct quotes and inline citations linked immediately back to the original source material. As always, it’s essential to consult your district’s technology team and legal counsel to ensure that all privacy concerns and safety risks have been reviewed.
How the Notebook System Works for SBOs
The most prominent AI/enterprise software systems all have some version of a source-grounded AI tool: Microsoft 365 has Copilot Notebooks, OpenAI has ChatGPT Projects, and Google has NotebookLM. Each allows the user to create a centralized archive of information, like an institutional memory, from which the AI will work.
For example, in NotebookLM, SBOs can upload documents, add links to websites, or add from their drive — up to 400 sources per Notebook — to create a powerful, data-grounded knowledge base. The examples below reference NotebookLM, but the principles can be applied to many similar tools from other companies.
To illustrate the practical application, imagine building a Long-Term Planning Notebook designed to project a whole host of district metrics, financial and otherwise, out into the future:
1. Gather Your Sources: Upload all relevant internal and external documents. These might include:
-
-
Demographic data from sources like the U.S. Labor Department, Census Bureau, and local sources for employment, births, and population trends.
2. Organize for Efficiency: Instead of uploading dozens of fragmented reports, consider using the collapsible document architecture. Using deep research, generate reports on sources that are on the same topic. You can then submit that report as a source, which helps consolidate long source lists and makes information easier to find.
3. Encourage Team Use: Share the Notebook so collaborators can add sources, bringing additional insights into the long-term planning process.
Actionable Prompts: Getting Practical Output
Once you have a notebook filled with quality sources, the fun can begin. You can customize personas in NotebookLM. In the “Configure” menu at the top of the chat box, you can use the preset characteristics or create your own. You could prompt it to be a creative strategist, looking for non-obvious connections and innovative applications of your data for long-term planning. Or you could ask it to serve as the skeptical reviewer, identifying inconsistencies and asking questions to refine your output documents.
Here are examples of prompts tailored for high-impact SBO tasks:
1. Function: Contract Review / Compliance
Example Prompt: "Compare the 2023 version of the teachers union contract against the new 2026 version, highlighting all substantive changes. Compare the new 2026 version to all the forms and procedures on our website to point out where we need to make updates."
Expected Outcome and Benefit: Rapidly identifies important changes for stakeholders, identifies documents needing updates to ensure the district stays compliant faster than manual review.
2. Function: Financial Reporting / Board Prep
Example Prompt: "Generate a concise briefing that summarizes the key updates from the last three Board presentations, focused on capital project spending, and highlight critical financial trends relevant for the upcoming meeting."
Expected Outcome and Benefit: Automates the collection and analysis of data over time, allowing the executive team to enter board discussions with a focused understanding of historical context.
3. Function: Strategic Scenario Planning
Example Prompt: "Assuming the Custom Persona of a 'CFO Advisor,' model arguments and generate three counter-perspectives regarding the proposed bus purchase based exclusively on the uploaded fleet policy and the manufacturer contracts."
Expected Outcome and Benefit: Supports robust, data-grounded scenario planning by modeling different strategic arguments based only on uploaded institutional data.
4. Function: Administrative / Onboarding
Example Prompt: "Generate a comprehensive FAQ document for new administrative assistants detailing the three most crucial departmental policy documents and referencing the sections where the policies are detailed."
Expected Outcome and Benefit: Automates the creation of essential training materials, significantly accelerating the onboarding process for new team members.
Building Notebooks and Exploring Insights
As a starting point, identify a topic or project you have in the short- to medium-term. Create a Notebook with all the internal sources you have: emails, meeting notes, quotes or contracts, audio recordings of meetings or calls, etc. The idea is to do a “data dump,” so all those disparate sources can be found in one place. NotebookLM can then be prompted to generate concise briefings, highlight critical trends, or suggest next steps, all with direct references to your sources.
You can also convert your AI-generated insights into new source documents, allowing the Notebook to use those notes when building future insights. NotebookLM provides options to generate complex outputs such as slide decks, infographics, organizational charts, and audio overviews directly from source material, so you can tailor your output to different audiences.
NotebookLM’s source-based methodology offers a compelling tool for the SBO office, turning mountains of siloed documentation into an active, reliable resource for compliance, planning, and knowledge transfer. The key is giving the tool high-quality, relevant sources and clear, actionable instructions.