I’ve written before about my Reed-Harrison brick wall (and my shady maternal 5th-great-grandfather) and how I’m using AI to help find my way around it. While I’ve learned a great deal—though no major breakthrough yet—I wanted to describe how I’ve been using Gemini’s NotebookLM (now simply Notebook) as part of that research.
If you already use Google services, getting started with Notebook is straightforward. The interface consists of three panes: Sources, Chat, and Studio. Sources contain your input—documents, spreadsheets, and other files. Chat is where you enter prompts and receive responses. Studio serves as something of a repository and a playground, providing different ways to organize and present your findings.
So how did I use this new tool? First, I consulted Gemini in a separate chat, and it walked me through the setup, beginning with the creation of my source documents. Spending time planning and organizing at the outset can make a research project much more manageable. One early decision was to keep all my documents and spreadsheets in a folder on Google Drive and to use Google's own applications—Docs and Sheets. That allowed me to edit or add information online and then synchronize those files with the versions uploaded into Notebook's Sources environment. In the end, I had a collection of Google Docs and Sheets, each beginning with a two-digit number for easy sorting, along with several three-generation pedigree reports that I had downloaded and converted to text.
Since my focus was John Reed, some of the files were named 00_John_Reed_Narrative, 01_John_Reed_Timeline, 04_Muskingum_County_Records, and 11_John_Reed_Censuses. The narrative established my research objectives and held miscellaneous information that didn't fit elsewhere. The timeline followed the format suggested in Amy Johnson Crow's 31 Days to Better Genealogy. The census file contained tabulated information from each census, including the head of household, every family member's name, and the values of real and personal property.
The records files proved to be the most valuable—and the place where Notebook really demonstrated its strengths. Here I entered transcripts that I had obtained through FamilySearch's Full-Text Search. To keep everything organized, I maintained a separate research log in Google Sheets that recorded search terms, locations, years searched, source information, descriptions, and a Case ID (001, 002, and so on). Each transcript in my records document began with a header containing the Case ID, source, and image numbers before I pasted in the transcript itself.
Once I had accumulated enough information in each file, I uploaded it to the Sources pane. Each upload takes a few moments while Notebook analyzes the new material. Then it was finally time to enter my first prompt into the Chat pane. Unlike a standard chatbot, which may draw on its training data, search the internet, or occasionally make confident but unsupported assertions, Notebook restricts itself to the sources you have provided.
What did it return? In my case, it identified family relationships and inheritances, untangled who was suing whom, traced land transactions, and explained the legal terminology found in nineteenth-century court records. It also disentangled complicated family relationships and highlighted inconsistencies among the various records, especially in pedigree files I had downloaded from FamilySearch.
It summarized what I had learned about my ancestor John Reed:
Through the analysis of court transcripts, probate records, and census data, a detailed portrait has emerged of John Reed (1793–1873), a man whose life in Muskingum and Morgan counties, Ohio, was defined by a complex web of family ties and relentless legal conflict.
It also identified the key documents supporting its conclusions. My John Reed project quickly became complicated, with connections that would once have required stacks of index cards, a box of pushpins, and yards of string to keep straight. Instead, Notebook organized the evidence, identified relationships, and documented the reasoning behind its conclusions.
[T]rying to track these "neighborly" interactions manually across 100+ images of transcripts and three different census years would be a "nightmare."
With NotebookLM, you were able to:
- Distinguish between separate lineages: It quickly separated the "Delaware Reeds" (William’s family) from your "Virginia Reeds," proving that the litigation wasn't a brotherly spat, but a neighborly exploitation.
- Track Economic Evolution: The tool allowed you to see John Reed’s real estate value jump from $3,000 in 1850 to $25,000 in 1870, providing the "motive" for why his neighbors and heirs were so litigious.
- Map the Migration: By cross-referencing witnesses like James Bay and boarders like John Lyder (both from Virginia), you used the "neighbor" data to reinforce the Loudoun County/Harrison connection—the very "brick wall" you started with.
One feature I particularly appreciate is that the analysis evolves as the project grows. As I discover new records, I simply add them to the appropriate Google Docs file and synchronize it with Notebook. When I found records relating to John's father-in-law in an adjacent county, I created another document, added it to the Sources pane, and let Notebook incorporate the new evidence into its ongoing analysis.
And what about the Studio pane? Each chat response can be pinned as a note, but Studio offers much more. Notebook can generate videos, slide presentations, data tables (which are especially useful for mapping FAN clubs), and reports. Other study tools, such as flashcards and quizzes, are also available, although those are probably of limited value for genealogical research.
So, while I can see only a little daylight through my brick wall, Notebook has allowed me to develop a much richer understanding of John Reed, his family, his neighbors, and the complicated world in which they lived. But I'm not done digging. Yet.







