Using Claude AI to Analyze Genealogy Research Reports
Do you follow the “leaders” exploring the intersection of genealogy and artificial intelligence? I do—and Facebook has become one of my favorite places to see how other researchers are experimenting with AI tools.
For some time now, the Genealogy and Artificial Intelligence Facebook group has been my go-to source for ideas, especially for prompt writing. Recently, however, another group began appearing regularly in my feed: Ancestors and Algorithms: AI for Genealogy. In one post, the group’s administrator described how he used Claude to organize and analyze historical facts drawn from genealogy research reports.
That immediately caught my attention.
Rather than asking Claude a single question or having it summarize one document, he fed the AI entire research reports and asked it to organize the information, identify patterns, and help make sense of a complicated research problem. Since I am currently working on a long-standing Crawford brick wall, I decided to see whether Claude could help me look at my data in a new way.
My Crawford FAN Club Experiment
My brick wall centers on James Crawford, who married Sarah Smith Duggins in 1799 in Garrard County, Kentucky. Sarah had previously been married to Alexander Duggins, and she had two sons—Henry and William—before Alexander’s death. The Crawford problem is complicated by the fact that multiple men named James Crawford were living in the same geographic area at roughly the same time.
To explore this, I created a FAN Club group in my RootsMagic file focused on Garrard County, Kentucky. That group included:
- Mary Crawford
- John Anderson
- James Crawford (wife Rebecca Anderson)
- James Crawford (wife Martha Knight)
- James Crawford (wife Sally Duggins) – my ancestor
- Moses Dooley
- Rebekah Douglass
- Thomas Kennedy
- Bazaleel Maxwell
- Alexander Moore
- James Sellers
- William Sellers
I then wrote a fairly lengthy prompt for Claude explaining the known relationships, migrations, and record overlaps among these individuals. This included background on Garrard County’s formation from Madison and Lincoln Counties, shared migration paths from Virginia into Kentucky, and the interconnections between the Crawford, Anderson, Maxwell, Douglas, Sellers, and Duggins families.
Feeding Claude the Research Reports
My original plan was to attach full research reports for everyone in the FAN Club and ask Claude to generate a comprehensive timeline. That quickly ran into a practical limitation: the reports were simply too long to attach all at once.
Instead, I narrowed the experiment to research reports for the three different men named James Crawford and added a focused instruction:
Attached are research reports for the three James Crawfords. Can you create a timeline concentrating on the time period of 1775 to 1820?
What Claude produced was surprisingly useful.
Claude’s Timeline Analysis
Claude created a side-by-side analytical timeline distinguishing the three James Crawford men by spouse, geography, military service, land records, tax lists, and migration paths. It clearly separated:
- James Crawford (husband of Sally Smith Duggins) – my ancestor, who migrated from Kentucky to Preble County, Ohio
- James Crawford (husband of Rebecca Anderson) – a Revolutionary War veteran who later moved to Jefferson County, Indiana
- James Crawford (husband of Martha Knight) – believed to be the son of Rebecca Douglas Crawford, with ties to Barren County, Kentucky, Preble County, Ohio, and later Warren County, Indiana
Claude also highlighted areas of geographic overlap, particularly in Madison and Garrard Counties during the late 1790s and early 1800s—exactly where much of the confusion arises in the records. It flagged potentially significant deeds, tax listings, and shared associates, and it summarized migration patterns in a way that made visual sense.
Most importantly, Claude did not claim certainty. Instead, it identified possibilities, conflicts, and areas where additional research is needed—precisely what a genealogist wants from an analytical assistant.
What Worked—and What Didn’t
This experiment reinforced something I’ve learned repeatedly while using AI for genealogy:
AI doesn’t solve brick walls—but it can help you think more clearly about them.
Claude didn’t discover new records or prove relationships. What it did do was:
- organize complex data quickly
- highlight overlapping timelines and locations
- surface patterns that were easy to miss when reading reports individually
- suggest productive avenues for further research
The biggest limitation was input size. Long, detailed research reports still have to be curated and selectively provided. That said, even a partial dataset produced valuable insights.
Final Thoughts
Using Claude to analyze genealogy research reports won’t replace careful record analysis, source citations, or original research. But as a thinking partner, it can be extremely helpful—especially when dealing with multiple same-name individuals, overlapping FAN clubs, and tangled migration paths.
For me, this was another reminder that AI tools are best used in partnership with traditional genealogy skills, not as a substitute for them. And sometimes, a fresh way of organizing what you already know is exactly what’s needed to move a brick wall forward.
📜