Ep. 33: FAN Club Method + AI - Find Ancestors Through Their Neighbors

If you have a brick wall ancestor with a common name; a William Harrison, a Mary Smith, a John Thomas this episode will change how you research forever.

In Episode 33 of Ancestors and Algorithms, host Brian walks through a completely upgraded FAN Club cluster research workflow that goes far beyond what was covered in Episode 16. The FAN Club method, coined by renowned genealogist Elizabeth Shown Mills, teaches researchers to find answers through their ancestor's Friends, Associates, and Neighbors when direct records fail. Combined with today's AI tools, it is one of the most powerful brick wall strategies available to family historians anywhere in the world.

This episode features four free AI tools; Claude, ChatGPT, NotebookLM, and Perplexity, each assigned to a specific role in a step-by-step research pipeline:

→ Perplexity researches the geographic migration corridors and record repositories tied to your ancestor's community 

→ ChatGPT builds a structured cluster analysis strategy before you search a single record 

→ Claude analyzes your extracted census neighborhood data to identify surname clusters, birthplace patterns, and priority FAN club members 

→ NotebookLM organizes all your research evidence into a single, source-grounded command center

You will hear a complete composite research case from start to finish, including how a neighbor from the same Indiana county led directly to the ancestor's Ohio origins; using only free tools and publicly available records.

Whether you research US census records, UK parish registers and tithe apportionments, or Australian colonial musters and land selection records, the FAN Club principle works in every record system. This episode explicitly addresses all three research traditions with actionable strategies for each.

The research methodology demonstrated aligns with the Genealogical Proof Standard and is referenced throughout the episode; showing how AI assists serious genealogy without replacing rigorous research practice.

What you will walk away with: three copy-paste ready prompts, a four-tool workflow you can use this week, and a new way of looking at every census page you have ever seen.

All prompts and resources from this episode are available free at ancestorsandai.com. Patreon members at ancestorsandai.com receive additional intermediate and advanced prompt guides built directly from this episode's research workflow.

Connect with Ancestors and Algorithms:

📧 Email: ancestorsandai@gmail.com
🌐 Website: https://ancestorsandai.com/
📘 Facebook Group: Ancestors and Algorithms: AI for Genealogy - www.facebook.com/groups/ancestorsandalgorithms/

Golden Rule Reminder: AI is your research assistant, not your researcher.

Join our Facebook group to share your AI genealogy breakthroughs, ask questions, and connect with fellow family historians who are embracing the future of genealogy research!

New episodes every Tuesday. Subscribe so you never miss the latest AI tools and techniques for family history research.

Transcript

SPEAKER_00 0:11

There are two kinds of genealogists. The ones who search for their ancestor by name, hit a wall, and stop. And the ones who flip the script look at everyone around their ancestor and find them hiding in plain sight. Back in episode 16, I taught you the fan club method using census records and a basic AI extraction workflow. Several hundred of you went home and tried it. You sent me emails. You posted in the Facebook group. And a lot of you said the same thing. Brian, I got the data, but I got lost in it. What do I actually do with a spreadsheet full of strangers? That's a fair question. And today I have a much better answer. Because in the four months since that episode, the AI landscape has changed in ways that make cluster research faster, smarter, and dramatically more powerful. And I want to show you what that looks like in practice. This is the episode I wished I had made four months ago. Let's get into it. Welcome to Ancestors and Algorithms, where family history meets artificial intelligence. I'm your host, Brian. Today we're doing something a little different. This is both a callback and a complete level up. If you listened to episode 16, you got a solid foundation in the fan club method, which stands for Friends, Associates, and Neighbors. That methodology was coined by Elizabeth Sean Mills, one of the most respected genealogists in America, and it is still one of the most powerful brick wall busting techniques in our entire toolkit. But since then, we've crossed into Patreon territory. And you, the people who are investing in this podcast, deserve a deeper, more sophisticated look at how to actually apply this methodology with today's AI tools. So today we're going back to school on cluster research. We're going to use a composite research scenario, the kind of common name brick wall that I guarantee most of you have hit at some point, and we're going to walk through a complete multi-tool workflow that is meaningfully upgraded from what I showed you in episode 16. New strategies, new capabilities, prompts you can copy and use today. Let's find some ancestors. I want to start by being honest with you about something. When I recorded episode 16, I was genuinely excited about the fan club method. I still am. But looking back at that episode, I realized I taught you the what without fully teaching you the what next. I showed you how to extract a census neighborhood into a spreadsheet, and I showed you how to find one specific person by spotting a pattern. What I didn't show you was what to do when that answer isn't that obvious. When you have sixty seven strangers in a spreadsheet and no clear candidate jumps out. That is the harder problem. And that is what today is about. So let me introduce you to a composite scenario that represents one of the most common brick walls I hear about from listeners and researchers everywhere. I call him William Harrison. Now I want you to think about every common name brick wall you have ever had. John Smith, Thomas Anderson, William Johnson. These are the ancestors who stopped research cold because there are simply too many of them to distinguish without surrounding context. William Harrison fits that mold perfectly. He shows up in the eighteen eighty federal census for Wayne County, Indiana, living with his wife Mabel and three young children ages two, four, and seven. His listed occupation is farmer. His birthplace is Ohio. His parents' birthplaces are listed as Ohio. That's it. That's everything the 1880 census gives you. Now, if you search for William Harrison born Ohio, farmer, Wayne County, Indiana, 1880, you are going to find approximately 47 William Harrisons who could plausibly match. And that's not an exaggeration. In the 1880 Midwest, William Harrison was one of the most common name combinations a person could have. The standard approach, checking vital records, searching earlier censuses forward, hits walls fast. Because without knowing his parents' names, his exact birth year, or his original Ohio township, you are searching in the dark. But here's where the fan club method changes everything. And here's where AI in 2026 changes the fan club method. You see, William Harrison doesn't exist in isolation. He lived in a community. He had neighbors who may have come from the same Ohio County he did. He had witnesses to his land transactions. He may have attended the same church as people he grew up with. His wife Mabel's family may be nearby. The people around William Harrison carry his origin story in their own records, even when his records won't cooperate. Elizabeth Sean Mills, who coined the fan principle, put it exactly right. Historical information is like real estate. The true value of any piece of information is unknown until it is put into community context. And I want to say this clearly for every listener, wherever you are in the world. That statement is as true for a family in rural Yorkshire as it is for a family in rural Indiana. It is as true for a convict settlement in New South Wales as it is for a Quaker township in Ohio. The fan club method is a human principle, not an American one. The records look different. The principle is identical. AI lets us put that community context together faster than ever before. But here's the thing I want you to hold on to as we work through today. The AI is going to help us organize, analyze, and strategize. That is genuinely powerful. But AI is your research assistant, not your researcher. Every lead it gives you still has to be followed up in primary records. Every pattern it spots still has to be verified by you. That doesn't diminish what AI can do. It just means we go into this as partners, not passengers. So let me walk you through the upgraded workflow. Four tools, four distinct roles, one strategy that goes well beyond a spreadsheet full of strangers. When I did episode 16, my workflow was essentially upload a census image, extract it with Gemini, paste it into a spreadsheet, and then sort and look for patterns. That is still a perfectly valid starting point for the data extraction step. Gemini in AI Studio remains excellent at handwriting transcription and that workflow holds up. But what I want to show you today is what happens after that extraction. Because that's where the real work is and that's where the newer AI capabilities have made a dramatic difference. So step one, you've extracted your census neighborhood using Gemini in AI Studio. You follow the process from episode 16, or you can find those prompts on the episode 16 companion guide. You have a spreadsheet with all of William Harrison's neighbors, names, ages, birthplaces, occupations, immigration years of applicable, parents' birthplaces. Now what? The first thing I want to do before I even start analyzing that spreadsheet is understand the historical and geographic context of the community. And for that, I turn to perplexity. If you haven't used perplexity, here's the short version. It's a research engine that searches the current web and gives you cited sourced answers. It's particularly good at questions like what records exist for this place and time? And it's free at the basic tier, which is all we need today. Here's the prompt I use for the William Harrison scenario. Quote, I am researching a family who lived in Wayne County, Indiana in eighteen eighty. The male head of household was born in Ohio around eighteen forty eight, and his parents were also born in Ohio. Wayne County, Indiana sits on the Ohio border. I need to understand what Ohio counties border Wayne County, Indiana? What were the most common migration corridors from Ohio into this part of Indiana in the eighteen fifties through eighteen seventies? And what record types in Wayne County, Indiana from eighteen seventy to nineteen hundred might help me identify which Ohio County a family originally came from. Please provide specific record repositories with links where possible, end quote. This is not a question I could answer efficiently from memory. And perplexity comes back with something genuinely useful. It pulls current information about Wayne County's geographic setting, identifies the two Ohio counties that directly border it, Preble to the southeast and Dark to the northeast, and notes that Preble County in particular was a documented feeder county for Wayne County, Indiana migration. It also points me to the Indiana State Archives, the Indiana Historical Society, and the Wayne County Historical Society as primary repositories. For my Australian listeners, the National Archives of Australia has comparable regional migration records for settlers from specific British counties, and the state archives in each state hold settlement era records worth exploring for the same principle. UK researchers can look to county record offices where parish records and overseers of the poorer records often document migration change in remarkable detail. That geographic context from perplexity is going to shape everything else I do. Now I know I should be paying special attention to neighbors who show birthplaces in those three Ohio counties because they may have come west in the same migration wave as William Harrison's family. Here's where today's episode parts ways from episode sixteen in a significant way. In episode sixteen, I was essentially looking at the extracted census data and using my own research intuition to spot patterns. My human eyes, my knowledge of genealogy, scanning for something useful. That works, but it's slow. And it depends entirely on what I already know to look for. In 2026, I'm using ChatGPT as my cluster mapping strategist before I even start sorting data. ChatGPT is my creative thinking partner. It's not going to tell me facts, but it is excellent at research planning, at thinking through what questions to ask, and at generating search frameworks I wouldn't have thought of on my own. Here's the prompt structure I want to give you. This is one of the most valuable things I will share today. Pay attention here because this prompt can adapt to almost any fan club scenario you're working on. Quote, I am applying the fan club method in genealogy to find the origins of a brick wall ancestor. His name is ancestor name, and I have identified approximately number households in his 1880 neighborhood in county state. The key things I know about him are born approximate year in state, parents also from state, occupation, occupation. Please help me build a structured cluster analysis framework. Specifically, what categories of relationship should I be looking for among his neighbors? What patterns in the data would suggest a shared origin? What types of documents outside of the census should I look for among his cluster of people? And what questions should I be asking about the cluster as a whole that I might not have thought to ask? What ChatGPT comes back with for this kind of prompt is a genuinely organized research framework. It's not making up facts. It's thinking strategically alongside me. For the William Harrison scenario, it flagged four cluster patterns worth watching. Surname clustering, the same surnames appearing multiple times in the neighborhood, shared birthplace clustering, multiple neighbors from the same Ohio county or region, age clustering among the adults that might suggest a shared migration moment, and land transaction witnesses who appear in multiple deeds. That last one, the land transaction witness pattern, was something I wouldn't have led with on my own. But it makes perfect sense. In the 1870s, 1890s Midwest, when you bought land, you asked someone you trusted to witness it. And that trusted person was often a neighbor, a church member, or a family friend from your old community back in Ohio. For UK listeners, your equivalent documents are deed poll witnesses, manorial courtroll entries, and overseers of highways lists, all of which record people who interacted regularly within a defined geographic community. For Australian listeners, look at land selection application witnesses and squatting license witnesses from the pastoral era. Different documents, same human behavior. When something mattered legally, you called someone you trusted from your community. So now I have two things. I have perplexities geographic and repository context. I have ChatGPT's clustered analysis framework, and I have a spreadsheet full of William Harrison's neighbors. Now it's time for the heavy lifting. This is where the real upgrade lives. And this is where I genuinely wish I had had this conversation in episode 16. Claude is, in my experience, the strongest tool for what comes next. Multi-document pattern analysis and correlation of evidence. What I want to do is take my extracted census neighborhood data, combine it with the framework from ChatGPT, and ask Claude to do something that would take me hours manually. I want Claude to find the patterns in the cluster. Now, a quick note on how to do this effectively. You can paste your spreadsheet data directly into Claude as text. If you've extracted it as a CSV, you can either paste the raw text or in the Claude.ai interface, attach the file directly. Claude handles both well. Here's the approach. I'm going to give you the core of this prompt and then walk you through what Claude does with it. Quote, I am conducting fan club cluster research in genealogy using the methodology developed by Elizabeth Sean Mills. My focus ancestor is William Harrison, born approximately 1848 in Ohio, living in Wayne County, Indiana in 1880 with his wife Mabel and three children. Below is the census data I have extracted for approximately 60 households in his immediate neighborhood. Paste your spreadsheet data here as text or attach the file. Please analyze this cluster of neighbors and do the following. One, identify any surname clusters, the same last name appearing in multiple households. Two, identify any birthplace clusters, multiple neighbors sharing the same state or region of origin. Three, identify any households that share the same Ohio birthplace as William Harrison or whose parents share Ohio origins. Four, note any neighbors whose ages and origin patterns might suggest they migrated to Indiana in the same wave as William Harrison. Five, create a short list of the top three to five households in this cluster that would be the highest priority for follow up research with your reasoning for each. Do not speculate about relationships, identify patterns only. I will do the genealogical interpretation end quote. That final instruction is important. Do not speculate about relationships, identify patterns only. That is how you keep AI in the assistant role rather than letting it start inventing connections. It is doing the data work. You are doing the genealogy work. For the William Harrison scenario, Claude comes back with something quite interesting. Among the sixty two extracted household, it flags three distinct surname clusters, two households with the surname Hively, three with Garretsen, and two with Pemberton. It notes that eight of the neighbors have parents listed as Ohio born, and three of those list their own birthplace as Ohio as well, a cluster worth investigating given what perplexity told us about Prebou and Dark Counties as primary migration sources in this part of Indiana. And then it creates the short list. The top priority, according to Claude's pattern analysis, is a household two doors down, an Ezra Gerritson, age fifty two, birthplace Ohio, parents birthplace Ohio. His wife is also listed as born in Ohio. Two of their adult children are in their twenties, suggesting this family planted in Indiana perhaps ten to fifteen years before the eighteen eighty census. Here's where I want to stop and explain what just happened. Because this is the moment that separates this workflow from episode sixteen. In episode sixteen, I was looking for a sibling. A specific person I already suspected was nearby. The pattern was obvious once I knew what to look for. Today we're not looking for a specific person. We're building a contextual web. And the Garretson family is now interesting to me not because they might be related to William Harrison, but because they might be from the same Ohio community. And if they are, Ezra Garrettson's records in Ohio will tell me something about where that Ohio community was, which then tells me where to look for William Harrison. That is the fan club method working at its fullest. You're not just finding relatives. You are reconstructing the community your ancestor lived in before the records found them. And this works in any record system. UK researchers, you do this with neighboring farms in a tithe apportionment, with tenants listed together in a rent roll, or with ratepayers clustered in the same township in a rate book. Australian researchers, you do it with passengers who arrived on the same assisted immigration ship and then took up land selections in the same district, or with convicts assigned to neighboring properties whose names appear together in bench books and mustard rolls. The records are different. The community bond that connects those names is exactly the same. This is where Notebook LM becomes critical. And I want to say something important about Notebook LM here. It is not a chatbot. It does not pull from the internet. It only works with the sources you give it. That limitation is actually what makes it powerful for this step. Because now I have multiple layers of information that all need to talk to each other. I have Perplexity's repository research. I have ChatGPT's analysis framework. I have Claude's pattern analysis from the census data. And I'm about to go pull actual records on Ezra Garrettson and the other prioritized neighbors. Nobook LM is my research command center. I upload all these working documents as sources. My extracted census data, my perplexity summary on Wayne County repositories, my Claude Pattern analysis, and then as I find new records on the prioritized neighbors, I add those too. What Notebook LM gives me is the ability to query across all of those sources at once and ask questions like, based on everything I've uploaded, what do we now know about the likely Ohio origin of William Harrison's neighborhood cluster? And it answers based only on what I've actually found. No hallucinations, no guessing, just synthesis of my verified research. Notebook LM has also added some genuinely useful features in recent months. The data table function is particularly good for this kind of work. You can ask Notebook LM to pull all of the key facts from your uploaded sources into a structured table. Names, dates, locations, source citations. It's a huge time saver for the kind of cross-referencing that cluster research demands. I want to pause here for a second and acknowledge something that I think about every time I work through a workflow like this. What we're doing right now is exactly what the genealogical proof standard demands of serious researchers. We're not just finding records. We are correlating evidence from multiple sources, analyzing the patterns, and building a case that would hold up to scrutiny. That is GPS elements one, two, and three all running at the same time. Exhaustive research across multiple record types, careful sourcing and documentation of every step, and thorough analysis and correlation of the evidence we find. AI is helping us do that faster, but the standard has not changed. And it should not change. So let me tell you what the Garretsen lead turned into. I searched ancestry and family search for Ezra Garrison in Ohio before eighteen seventy, Preble County Land Records, the eighteen sixty and eighteen seventy censuses, and there he is. Ezra Garrison in the eighteen seventy census, still in Preble County, Ohio, living on a farm, and three households away on that same eighteen seventy Preble County Census page, a George W. Harrison, age fifty two, born Ohio, farmer, with a son listed, William, age twenty two, born Ohio. William Harrison, age twenty two in eighteen seventy, born Ohio, living in Preblo County, Ohio. Do you see what just happened? The Garretson family, who followed William Harrison to Indiana, left a trail in Ohio that led back to William Harrison's own family in Ohio. We didn't find William by searching for William. We found him by searching for his neighbor. That is the fan club method. And that is the power of doing it with AI assisted cluster analysis. I want to be clear, this is a composite scenario designed to illustrate the workflow. But this exact sequence of events, neighbor leads to Ohio County, Ohio County leads to ancestor, happens in real research all the time. I've seen it. Members of our Facebook group have shared stories that follow almost this exact pattern. Here's something I want to highlight because it speaks directly to what has changed since episode 16. In that episode, the process was essentially manual pattern recognition. You extracted the census page, you sorted the spreadsheet, you used your own eyes and your own knowledge to find the person who looked interesting. That works if the answer is obvious. If the sibling is two doors down and the birthplace matches and the surname is identical, you'll find it. But what if the connection is less direct? What if it's not a sibling but a neighbor from the same Ohio township who has completely different surname? What if the migration pattern is the clue, not the names? Your eyes and your intuition can miss that. Claude doesn't. When you give Claude sixty rows of data and ask it to find clusters, it processes every single row with equal attention. It does not get tired. It doesn't skip the Gertson household because the name's different from Harrison. It finds the pattern because you asked it to find patterns. That is the specific capability that's been transformative for this methodology in 2026. And I think it's worth naming clearly because it explains why combining AI with the fan club method is not just faster, it's deeper. You are accessing a layer of the cluster that human only analysis often misses. Now, I said we would talk about what happens when it doesn't all line up neatly, because it rarely does. In the William Harrison scenario, when I searched for George W. Harrison in Preble County, Ohio in the eighteen seventy census, I actually found two of them. Two George W. Harrisons, both with a son named William, both in neighboring townships. This is where Claude earns its keep again. I took both census records, transcribed them, and asked Claude to help me determine which was more likely to be my Williams family. I gave Claude the eighteen eighty Indiana data on William Harrison, the two candidate eighteen seventy Ohio households, and asked it to analyze the evidence. Claude looked at the ages, the neighbors, the farm values, the birthplaces of the children, and it flagged something I had not noticed on my own. In the eighteen eighty Indiana census, William Harrison's oldest child, age seven, was listed as born in Ohio, which means William was still in Ohio at least until eighteen seventy three. One of the two George W. Harrison candidates had already moved away from Puebla County by 1872, according to the County History Records and Perplexities Research Summary. That eliminated one candidate. Not with certainty, with evidence. And that's what matters. This is one of my favorite things to say on this podcast because it genuinely matters. AI is your research assistant, not your researcher. Claude didn't make the genealogical judgment call that eliminated candidate number two. It organized the evidence. I made the call. That is the correct division of labor. So where does this leave us? We now have strong circumstantial evidence that William Harrison of Wayne County, Indiana was the son of George W. Harrison of Prebla County, Ohio. We got there not by finding William Harrison's records directly, but by tracking one of his Indiana neighbors, Ezra Garrison, back to his Ohio origins and discovering that both families appear on the same Ohio census page in 1870. That is what Elizabeth Sean Mills means when she says your ancestors fan club are often the key to questions your direct research can't answer. We were stuck on William, but the community William belonged to was not stuck. It was findable, and it led us back to him. Now, before I wrap this research up and call it a conclusion, I need to do what every serious genealogist has to do. I need to verify this in additional sources. A pattern in census records is a lead, it is not proof. So the next steps for this research are clear. First, I search for probate records in Preblock County for George W. Harrison. If George died in Ohio and named a son William in his will, that's corroborating evidence. Second, I look for church records in the township where this Harrison family lived. Third, I search for William and Mabel Harrison's marriage record. Mabel's maiden name may appear there, and her parents' origins may add another piece to the cluster. Perplexity has already given me a list of where those records are. The Preblo County Probate Court Records, the Ohio History Connection in Columbus, Quaker meeting records for that region of Ohio and Indiana, both Wayne County, Indiana and Preblick County, Ohio had active friends meetings in the mid-1800s, and the Quakers kept detailed membership and removal records that are excellent for tracing migration, and the Wayne County, Indiana Marriage Register for the early 1880s. That is a research plan. Built in one session using four AI tools, grounded in a methodology that professional genealogists have used for decades. Here's what I want you to take away from today that goes beyond the William Harrison scenario. The fan club method is not a trick. It is not a workaround. It is a fundamental principle of how communities worked and how records reflect those communities. People migrated together. They signed each other's deeds. They were godparents to each other's children. They sat on the same church benches. They witnessed each other's wills. They wrote letters home about the good land available in a new county. And then a brother in law followed. And then a cousin. And then a neighbor from two farms over who had always said he'd come west someday. That pattern played out on every continent. In the UK, entire Cornish mining communities followed the copper seams to South Australia, arriving within a few years of each other and settling in clusters around Borough and Capunda. Scottish crofters displaced by the clearances, arrived in Victoria in groups and took up neighboring selections. In England, entire villages from specific parishes in Devon or Dorset appeared together on the same assisted passion ships because a local landowner sponsored them. The details differ. The impulse is identical. Go where people you know have already gone. That community migration pattern is your research superpower when your ancestor is invisible in direct records. Because the community members weren't all invisible. What AI gives you in 2026 in a way it simply didn't four months ago is the ability to process that community data quickly and systematically. Claude can analyze 60 households in seconds. Notebook LM can correlate evidence across a dozen uploaded documents. Perplexity can tell you exactly which repositories hold the records you need. ChatGPT can build your research strategy before you start looking. People ask me, Brian, doesn't this feel like cheating? Like AI is doing the research for me? I want to push back on that firmly. Ezra Gertson did not become interesting because Claude found him. He became interesting because I asked the right question, interpreted the pattern Claude returned, made the decision to follow that lead, and then searched ancestry and family search with my own hands and connected the dots. Claude organized the data so I could think more clearly. It didn't think for me. That's not cheating. That is using a tool. And genealogists have always used the best tools available to them. And through all of it, every single step, the underlining standard stays the same. AI is your research assistant, not your researcher. You are the genealogist. The tools make you faster and more organized. But the judgment, the verification, the proof, that belongs to you. This is what the genealogical proof standard asks of us. Reason exhaustively, cite everything, analyze and correlate the evidence, resolve the conflicts, and write a soundly reasoned conclusion. AI can help with all five of those steps, but it cannot substitute for any of them. And for those of you who want to see what that final written conclusion step looks like with AI as your drafting partner, that is exactly what episode 42 is going to cover. We are building toward a complete GPS methodology series, and the capstone episode on writing proof argument is going to be a good one. Stay with us. Here's your homework for this week. I'm giving you three levels, and this week I want to be explicit. These levels apply no matter where in the world you are researching. Cluster research is not an American methodology, it is a human methodology. Communities in England, Scotland, Ireland, Australia, and every other corner of the world operated on exactly the same principles. People moved together. They settled near people they knew. They signed each other's documents. The fan club principle works wherever your ancestors left records. Beginner level, take any record for one of your ancestors that shows their geographic community. For US researchers, that is a census record. For UK researchers, it might be a parish register, a tithe apportionment, or a set of poor law records. For Australian researchers, it could be a colonial muster, a land selection record, or a rate book. Extract the surrounding 15 to 20 households or entries using the Gemini and AI Studio workflow from episode 16. Then paste that data into Claude and ask it to find surname and birthplace clusters. See what patterns emerge. Intermediate level, run the perplexity context step first before you touch any records. Ask perplexity about the migration patterns relevant to your ancestors' community. US researchers, you know the drill. UK researchers, ask perplexity which English, Scottish, Welsh, or Irish counties fed into your ancestors' parish or township and what records survive at the National Archives at Q or at the relevant county record office. Australian researchers, ask perplexity about the ship arrival patterns and settlement corridors for your colony and era. The geographic context shapes everything that comes after. Advanced level, run the complete five tool workflow. Perplexity for historical and geographic context, ChatGPT to build your cluster analysis strategy, Gemini and AI Studio for extraction, CLOD for pattern analysis, and Notebook LM to hold everything together. Pick one brick wall ancestor with a common name. Build the cluster. Follow the highest priority neighbor lead into at least one additional record type beyond where you started. US researchers follow that neighbor into land records, probate, or church records. UK researchers follow a neighbor from the parish register into the census, into nonconformist records, or into the freeholders list. Australian researchers, follow a neighbor from the muster into the land grants, the bench books, or the assisted immigration records. See where it takes you. And I want to hear from all of you, not just the Americans. The whole point of this community is what we learn from each other's experiences with different record systems. A UK listener finding a cluster connection through tithe apportionments or an Australian listener tracing a convicts neighbor network through assignment records, those stories are genuinely instructive for everyone in this community. Come find us at ancestorsandai.com. Post what you find in the Facebook group. Tell us who you found through someone else's records. Those are my favorite posts. If you want to go deeper than today's episode took you, the companion guide for episode thirty three is available to Patreon members at ancestorsandai.com. It includes twelve advanced prompts covering multigenerational cluster analysis, AI assisted probate research for fan club members, template frameworks for documenting your cluster in a way that meets GPS standards, and a step-by-step workflow for connecting Burn County Ancestors through their neighbors' surviving records. Today's episode gives you everything you need to get started. The companion guide is for when you're ready to take this to the next level. Before we wrap up, if this episode taught you something new or if episode 16 started you on a fan club journey that has led somewhere, please leave a review wherever you listen to podcasts and share it with a fellow genealogist. The people who would benefit most from this show are the ones who haven't found it yet. You can help them find it. We've covered a lot of ground today. The fan club method, the cluster research upgrade, four tools working together in a way that makes this methodology more powerful than ever. And through all of it, the same truth holds. You are the genealogist. The tools work for you. Next week on Ancestors and Algorithms, we are heading west, as far west as you can get here in America. Gold Rush, California, eighteen forty nine. I've got an ancestor who the family has always called a legendary forty niner, a man who went to California and struck it rich. Except when I went looking for him in the mining records, he wasn't there. Not in Tulum County, not in El Dorado County, not anywhere in the goldfields. The man was a ghost. And tracking down where he actually was and what he was actually doing took four AI tools, a degraded microfilm deed, and one very surprising county history from 1880. That is episode thirty four, the Gold Rush Ghost. And I promise you the ending is not what you expect. I'm your host Brian, and I will see you next week for another journey into the past powered by the future. Until then, happy researching.