Ancestors and Algorithms, Salt Lake City, UT (2026)

05/05/2026

New episode out today. Episode 36: The Highland Line: Tracing Scottish Ancestors with AI.

If you've got Scottish ancestors, especially Highland or Island ancestors, this is the episode I wish I'd had ten years ago. We're walking through the exact AI workflow for Scottish research, from the orientation that saves you days of confusion, to the AI that reads handwritten 1810 marriage proclamations like they were typed yesterday, to the cross-document analysis that turns scattered census entries into a coherent family story.

The case at the center: a woman who shows up in the 1851 Inverness census as Catherine MacLeod Gillies, born on the Isle of Skye. Three AI tools later, we had probable parents, an explanation for why the direct proof is missing, and a very specific list of exactly what to search next.

Three AI tools featured. Three copy-paste ready prompts. One honest Partial Answer that demonstrates exactly what good genealogical research actually looks like.

05/04/2026

May the Fourth be with you, genealogists.

You know what always gets me about Star Wars? Luke Skywalker spent years not knowing who his father was. He had no idea about his family's history, where he came from, or why certain things about him felt so unexplainably familiar. And then one conversation changed everything.

Sound familiar?

That's genealogy. That's the moment you pull a record you weren't expecting, and suddenly a story you thought you knew looks completely different. The Skywalkers had Darth Vader. Your family has secrets too, and some of them are just waiting in an archive somewhere.

Vader literally means "father" in Dutch. George Lucas knew what he was doing.

Happy researching, and may the Force (and the records) be with you.

05/01/2026

There's a meaningful difference between explaining a conflict away and actually resolving it.

Explaining away says: "That record is wrong."

Resolving it says: "Here is why that record says what it says, and here is what it tells us about this person's life."

GPS Element 4 is the difference between a conclusion that holds up and one that collapses the moment someone looks closely.

Episode 35 walks through that distinction with three AI tools and a real case. Link in the first comment.

04/30/2026

The three-step AI workflow from Episode 35 applies to any conflicting records problem you're working through right now.

Step 1 goes to Perplexity. Before you analyze a single record, research the historical context for the record types and the population your ancestor belonged to. What did these records typically capture? What are the known patterns of error for this era and location? That context is what lets you evaluate reliability intelligently.

Step 2 goes to Claude. Systematic document comparison, with one critical rule: tell Claude not to give you the conclusion. Ask Claude to evaluate each record's reliability for the specific field in conflict, identify who provided the information, whether they were present at the event, and what factors affect accuracy. The conclusion has to be yours.

Step 3 goes to ChatGPT. Before you narrow down to an answer, get every plausible explanation on the table. ChatGPT ranked the explanations for Conrad's church register entry by how well they fit the specific circumstances of his life, and the highest-ranked answer was the most human one.

Then the fourth step is yours alone: weigh the evidence, write the conclusion, and document the reasoning. That step cannot be delegated to any tool.

Every prompt is in Episode 35, copy-paste ready. Link in the first comment.

04/28/2026

Three records. Three different countries. One ancestor.

Conrad Becker left the Rhine Province of Prussia in the 1860s, arrived through the port of New York, spent years building his life in Kleindeutschland, and eventually settled in the coal fields of Schuylkill County, Pennsylvania.

His records can't agree on where he was born.

His 1875 church marriage register says New York. His 1880 census says Germany. His 1921 death certificate says Pennsylvania.

Here's what kept me up at night: one of those answers came from Conrad himself, spoken directly to his own pastor, recorded in German, in the pastor's own hand.

Episode 35 is live now. It walks through how three AI tools build a professional-quality resolution for conflicting records, not a guess, not a ranking exercise, but a defensible conclusion that accounts for every contradictory record using the Genealogical Proof Standard.

This is GPS Mini-Series, Part 2 of 4. If you've got a brick wall with conflicting records sitting in your own research, this one's for you.

Listen wherever you get your podcasts, or head to the website. Link in the first comment.

04/21/2026

For 170 years, one family believed their ancestor was a legendary forty-niner who went to California in 1849 and struck it rich.

Then I went looking for him in the records.

No mining claims. No county tax lists with "miner" written next to his name. Not in Tuolumne County. Not in El Dorado County. Not in Calaveras County. After going through every surviving Gold Rush county record I could find, the man who was supposed to be a family legend had completely vanished.

Episode 34 of Ancestors and Algorithms is live. "The Gold Rush Ghost."

This week I took that brick wall to four AI tools: Perplexity, Claude, ChatGPT, and Gemini via AI Studio. What came back rewrote the family story entirely. Not because the legend was wrong. Because the truth turned out to be a better story.

Here's what this episode covers:

-- Why Gold Rush records are so incomplete, and what actually survives from 1849 to 1860
-- How a nine-year age gap in two census records proved I had the wrong William Hartfield
-- The record category I'd completely overlooked, which cracked the case open
-- How four AI tools work in sequence, each doing what it does best
-- What a "research pivot" really means, and why finding an unexpected answer is often better than finding the expected one

Four prompts. All free tools. One story I didn't see coming.

Listen at the Ancestors and Algorithms website or search "Ancestors and Algorithms" wherever you get your podcasts.

[Link in first comment]

Know a genealogist who's been chasing a family legend they can't prove? Share this with them. This episode was made for exactly that.

04/19/2026

In case Episode 33 of Ancestors and Algorithms got buried in your feed this week:

"The Cluster Research Revolution: Finding Ancestors Through Their Neighbors"

The short version: if you have a common-name ancestor who's invisible in direct records, this episode shows you how to find them through the people around them, using a four-tool AI workflow built on one of the most powerful brick-wall-busting techniques in professional genealogy.

Three free prompts in the Facebook Group. Four tools, all free at the basic tier. About 30 minutes of your time.

Links to the episode are in the comments or find Episode 33 wherever you get your podcasts. Happy researching!

04/16/2026

Here's the thing about cluster research that most AI genealogy content skips over.

Using AI to analyze census data is not new. You can ask any chatbot to look at a list of names and find patterns. That part is table stakes.

What Episode 33 of Ancestors and Algorithms demonstrates is something more specific: a four-tool workflow built on a methodology that professional genealogists have used for decades, the FAN Club method developed by Elizabeth Shown Mills, applied in a way that makes it faster, more systematic, and accessible to researchers at every experience level.

Here's why the four-tool sequence matters. Perplexity goes first, not to look at your ancestor's data, but to understand the geographic and migration history of their community before you look at a single record. That context shapes everything. ChatGPT builds a research strategy framework before any analysis begins, including pointing to record types and patterns you wouldn't have thought to look for. Claude processes the full census neighborhood at once and surfaces patterns that human attention misses when scanning sixty rows of handwritten data. NotebookLM holds all the evidence together across multiple research sessions so nothing gets lost.

Each tool has a specific job. Together they make the methodology work in a way that manual research alone can't match.

All three prompts from Episode 33 are free in the Facebook Group. All four tools are available at no cost.

And the workflow applies whether you're researching in Indiana in 1880, in Yorkshire in 1851, or in New South Wales in 1828.

Head to Ancestors and Algorithms website for the full episode and information. Link in the first comment.

04/14/2026

New episode of Ancestors and Algorithms is live.

Episode 33: "The Cluster Research Revolution: Finding Ancestors Through Their Neighbors"

Here's the problem this episode solves.

Your ancestor has a name shared by forty-seven other people in the same state at the same time. You've checked vital records, searched census after census, and come up empty. Without knowing their parents' names, their exact birth county, or their original township, you're searching in the dark.

The FAN Club method, developed by genealogist Elizabeth Shown Mills, reframes the entire problem. Instead of searching for your ancestor directly, you research the people around them. The neighbors who may have traveled from the same Ohio county. The friends who witnessed their land transactions. The church community that moved west together. Their records carry your ancestor's origin story, even when your ancestor's own records won't cooperate.

In Episode 33, four AI tools work together to apply this methodology to a William Harrison scenario: a composite brick-wall ancestor with one of the most common name combinations in 1880 Midwest genealogy. By the end of the episode, we have strong circumstantial evidence of his Ohio origins — found not through his records, but through his neighbor's.

Three free, copy-paste ready AI prompts are in the Facebook Group. All four tools used in the episode are free at the basic tier. The episode is about 30 minutes.

This methodology works whether you're researching in Indiana in 1880, Yorkshire in 1851, or New South Wales in 1828. People migrated together everywhere. The records look different. The principle is identical.

Listen wherever you get your podcasts, or head to our Facebook Group for the prompts. Link in the first comment.

If you know a genealogist who has been stuck on a common-name ancestor, this is the episode to share with them.

04/11/2026

Still haven’t listened to Episode 32? Here’s what’s sitting at ancestorsandai.com waiting for you.

Episode 32 of Ancestors and Algorithms is the Irish genealogy episode. And it answers the question that stops more researchers than almost any other in family history:

What do you actually do when you hit the Irish wall?

Here’s the short version of what you’ll walk away with after 30 minutes:

A clear understanding of what the 1922 Four Courts fire actually destroyed, and more importantly, what it did not destroy. Because “all the records burned” is not accurate. Significant records survive. Knowing which ones is the first skill of Irish genealogy research.

A step-by-step framework using Perplexity and Claude to map surviving records for your ancestor’s specific Irish county, narrow a large set of Griffith’s Valuation results to a probable family cluster, and correlate your evidence into a GPS-standard research log.

Three copy-paste prompts you can use immediately on your own Irish ancestry research.

And awareness of one of the most important events in Irish genealogy in the last fifteen years: the 1926 Irish Census goes live on April 18th, 2026. Free. Fully searchable. For anyone with Famine-era Irish ancestry, this census could be the corroboration link between the ancestor who emigrated and the family they left behind.

If your family came from Ireland, even if you’ve always believed the records were gone, this is the episode to listen to.

Everything you need to start is in the episode. Everything you need to go deeper is at ancestorsandai.com.

04/08/2026

Episode 32 is live: “Across the Irish Sea: An AI Guide to Irish Research” 🍀

If you’ve ever tried to research Irish genealogy and hit the wall, you know exactly what I mean by “the wall.”

Someone tells you that all the Irish records burned in the 1922 fire. You look it up. You find out it’s true. And suddenly an ancestor you’ve been chasing all the way back to County Clare, County Cork, County Mayo, or anywhere else in Ireland seems completely out of reach.

Here’s what Episode 32 of Ancestors and Algorithms is about: that wall has more doors than you think.

In this episode, I trace one of my own ancestors, Caitlin Flanagan, from Boston, Massachusetts, in 1850 back to a specific townland in County Clare, Ireland, using two AI tools and entirely free genealogy resources.

Along the way, I use Perplexity to map the complete surviving record landscape for County Clare genealogy research in the Famine era, including what records exist, what’s been digitized, and what’s free. Then I use Claude to analyze 87 Griffith’s Valuation entries and narrow them down to a single candidate family in a single townland, using evidence from both the 1855 Griffith’s Valuation and the 1826 Tithe Applotment Books.

The result is a strong, qualified research hypothesis about exactly which Irish family Caitlin Flanagan came from. Built from land records, Catholic parish registers, and a GPS-integrated evidence correlation process.

This episode is specifically designed for:
Researchers with County Clare ancestry, County Clare genealogy research, Irish Famine ancestors, Griffith’s Valuation research, Catholic parish registers, and anyone who has ever been stopped by “all the records burned.”

And the timing is historic: The 1926 Irish Census, the first census of the Irish Free State, goes live on April 18th, 2026. Episode 32 gives you the research foundation to use it effectively when it drops.

Three copy-paste AI prompts included. All resources referenced are free.

Ancestors and Algorithms

05/05/2026

05/04/2026

05/01/2026

04/30/2026

04/28/2026

04/21/2026

04/19/2026

04/16/2026

04/14/2026

04/11/2026

04/08/2026

Address

Website

Alerts

Shortcuts

Share

Category