RLP 402: Transforming Court Records Research with AI: Finding and Logging

Nicole (1s):
This is Research Like a Pro, episode 402: Transforming Court Records Research with AI: Finding and Logging. Welcome to Research Like a Pro a Genealogy Podcast about taking your research to the next level, hosted by Nicole Dyer and Diana Elder accredited genealogy professional. Diana and Nicole are the mother-daughter team at FamilyLocket.com and the authors of Research Like a Pro A Genealogist Guide. With Robin Wirthlin they also co-authored the companion volume, Research Like a Pro with DNA. Join Diana and Nicole as they discuss how to stay organized, make progress in their research and solve difficult cases.

Nicole (41s):
Let’s go. This episode is sponsored by Newspapers.com. Hello everybody, welcome to Research Like a Pro.

Diana (50s):
Hi Nicole. How is it going today?

Nicole (53s):
Great. Today I discovered something on Ancestry DNA that was fun. I think it came out like eight months ago or nine months ago, but when you connect a DNA match to your Ancestry tree, it now says DNA citation created. And then if you go look at their profile in your family tree, it has a little source there in the list of connected sources. Like usually you’ll see a census record or a social security death index. Well now it has another one where it says like, DNA match to Nicole Dyer. And then So I added it to my grandma and now it shows that it’s a match between her and me and how many centimorgans we share.

Nicole (1m 41s):
And so that was cool. And I immediately had questions about privacy because I do have like a public tree, but I realized most of my matches are living. So they’ll all be listed as living in my tree if I add them. But then I thought, what about my grandma who’s deceased? And then I did connect a match with my grandma who’s deceased and then it was visible. So just be careful connecting matches to deceased matches to yourself ’cause then anyone who looks at their profile will be able to see how much they share with you.

Diana (2m 13s):
I’m so glad you brought that up because I haven’t done that or seen that yet. So now that we’re working with the DNA study group and I’m actively working on a new project for Cynthia Dillard using DNA to try to figure out her father who will be doing much more of that connection. So that’s great. Well let’s do some announcements. i. Have you ever hit a brick wall because you can’t figure out which John Smith is YOUR John Smith? You’re not alone. Genealogists everywhere struggle with the same frustrating problem: multiple people with the same name in the same place and time period. That’s exactly why Jan Joyce created “Merging & Separating Identities” – a multi-week intensive course that teaches you proven strategies to solve your most complex identity puzzles.

Diana (2m 59s):
Whether you’re dealing with many Henry Dewitts in one Ohio county or tracking someone who changed their name across multiple states, this course gives you the tools to crack the case. You’ll learn to build identity dossiers, master correlation techniques, and use both high-tech and offline methods to distinguish between candidates. Work with real case studies, practice with a common dataset, and get hands-on guidance from Jan and this team of expert instructors – Kristin Britanik, Nicole Dyer, Patti Hobbs, Lynn Nelson, and Kim Richardson. Stop spinning your wheels. Limited spots available. Visit the FamilyLocket website and click on Shop, then Institute Courses, to learn more and register for this dynamic “Merging & Separating Identities”

Diana (3m 42s):
course and finally break through those identity roadblocks. This institute course will be held weekly on Thursdays from April 30th to June 11th at 9:00 AM to 2:30 PM Mountain Time. Registration is closing soon, so be sure to register in the next week. Our Research Like a Pro Webinar Series continues and our April webinar will be on a Tuesday, April 21st at 11:00 AM Mountain time. Our presenter is Scott Dixon and he is teaching us about finding a father for Elizabeth Adcock. v. Elizabeth Adcock was born about 1786 in Granville County, North Carolina. In 1805, she married Henry Vincent and the couple migrated to Rutherford County, Tennessee along with a number of relatives and associates.

Diana (4m 30s):
So far, no direct evidence has been located to definitively identify Elizabeth’s father and family of origin. But ample documentary indirect evidence within the broader family and family network can help to identify her likely father. This case study makes strong use of locality, indirect evidence, and negative evidence to support its conclusion. Topics are Granville County, North Carolina; Rutherford County, Tennessee, Wills & Estates, Tax Records, Census, Land Deeds, Indirect Evidence, Negative Evidence. A little bit about Scott: has been seriously pursuing genealogical research since the late 1980s when he inherited several family Bibles and many letters, including ones from an ancestor couple who went to Africa as missionaries in 1856.

Diana (5m 16s):
His research has focused primarily on the Southeastern U.S., with forays into early Massachusetts and French Canada on his wife’s side. He is a member of the board of the Georgia Genealogical Society and three-time graduate of the RLP study group. Scott works in IT sales and lives in the beautiful Charleston, South Carolina area. Well, we are excited for that webinar. It is so great to learn from Scott. Our next study group begins August, 2026, and this is Research Like a Pro, without DNA. So a good, a good place to start. The peer group leader application is on our website if you’re interested in complimentary registration.

Diana (5m 56s):
Please join our newsletter that comes out every Monday to learn about new posts, our upcoming lectures to receive coupon codes. And we are looking forward to the next big conference, which is the National Genealogical Society Conference, May 27th through 30th at Fort Wayne. And this is where the renowned genealogy center at the Allen County Public Library is located. So we are so excited to go and get to visit that library. I believe it’s the second largest library after the Family History Library in Salt Lake City that has, you know, genealogy collections. So the theme for this conference is America at 250 and Nicole and I are both teaching.

Diana (6m 41s):
I am teaching about court records and female networks in one presentation and another one about uncovering family stories through court records. So kind of fun to have two on that topic. And Nicole is teaching about using AI to understand 250 years of historical terminology. And then a case study about DNA revealing Barsheba Tharp’s ancestry, and then a workshop for early handwriting. So lots of fun topics, not only from us but from all the other presenters, so we hope you’ll, you’ll meet us there.

Nicole (7m 16s):
Yes, I hope so too. And there are gonna be so many fun presenters. I always love attending this conference.

Diana (7m 23s):
Agreed. Well what are we talking about today? Speaking of court records, we are going to talk about court records and I think as Genealogists we always know that court records could be very, very helpful, but they are some of the hardest records to access, or at least they have been in the past. Part of the problem is, unlike vital records or census enumerations, they don’t really come with indexes. And so usually in the past you’d have to browse page by page through microfilm or the books. If you’re on site, you’re going to have to try to decipher the cryptic clerk notes, understand the archaic legal terminology, and piece together those narratives that can span multiple documents over several years.

Diana (8m 12s):
But what if you could search every word in thousands of a handwritten court pages in seconds? And this is where our AI assistant can really help us. We can have help finding the records as well as explaining some of those tricky legal terms or even trying to organize a multi-year litigation into a coherent timeline. So these are available right now and we are going to talk all about this as we go through a real research example, which was my latest little 14 day challenge project on identifying Samuel Daniel of Middlesex County, Virginia.

Diana (8m 52s):
And he’s the one we’ve talked about previously who advertised for a runaway apprentice, John Royston in 1770. And I believe it was on the podcast where we were talking about this, where I decided I needed to research Samuel Daniel. I had never done that. I know that sounds strange, I haven’t thought to do that, but I’ve been off researching other families. So I came back to John Royston and searched for Mr. Daniel. So my objective was, Research Samuel Daniel of Virginia who advertised for his runaway apprentice John Royston in the Virginia Gazette in 1770. Samuel Daniel was likely born before 1740 in Virginia. John Royston was born in 1750 in Gloucester County, Virginia, and died after 1816 in Georgia.

Nicole (9m 38s):
Good job on your research objective. It’s such an interesting one too. Most Genealogists would consider that as just as something interesting to do as a kind of a rabbit hole, like to research this other person who’s not your main person. That’s interesting. Well, the starting point was simple. It was that February, 1770 newspaper advertisement in the Virginia Gazette where Samuel Daniel was looking for his runaway apprentice, John Royston. Beyond that ad, nothing was known about Daniel. So you’re getting a fresh start on a person you’ve never searched on before, which is kind of fun because a lot of our research subjects we’ve already researched before a lot. So this was a fun fresh start, but you did have questions.

Nicole (10m 19s):
Was this the right county? What was his occupation? What happened to him? The answers were probably buried somewhere in the Middlesex County’s court records, but where traditional research would have meant systematically reviewing court records page by page, hoping to spot Daniel’s name among hundreds of other cases. The breakthrough came from a different approach entirely, and that is FamilySearch Full Text Search, AI powered. So that tool powered by artificial intelligence has been amazing and it reads handwritten documents and makes every word searchable. And I will say every word searchable, but every word that it can read correctly searchable.

Nicole (11m 4s):
That’s right. Instead of manually scanning digitized images, Diana can then search for Samuel Daniel across the relevant Middlesex County court records. In the FamilySearch catalog, she went to the page for Middlesex County, Virginia and then selected court records. The catalog shows 29 sources, but she narrowed it by date. She wanted the original court order books and found a great collection of them all digitized and available for Full Text Search. And probably some of those 29 were derivatives and other, you know, abstracts and just, there’s a lot of court records for a county that’s been along that time. Well Diana found two collections that she was the most interested in, given that John Roy’s disappearance occurred in 1770.

Nicole (11m 49s):
She wanted to search both before and after the date. So by clicking on the notebook icon on the far right with the AI symbol, you can go directly to that collection and use Full Text Search only within that digitized microfilm or collection. And this is very beneficial in narrowing down the search. We’ve talked about this before, it’s such a good way to know that you’re searching in the right place and time. You can also note the image group number or IGN and use it in the main Full Text Search window to narrow the results. The AI technology identified every page where Samuel Daniel appeared highlighting the relevant passages and the results were remarkable. 12 separate court record entries spanning April, 1773 through March, 1774, documenting a complex legal dispute between Samuel Daniel and John Royston and Richard Wyatt Royston.

Nicole (12m 43s):
So that was exciting because hey you are just researching a side person, but hey, here’s some court records about our ancestors that you hadn’t seen before because it’s in a county you hadn’t looked in. These records would’ve been more time consuming, to find through traditional browsing. They were appeared scattered across multiple court sessions over two years. These order books do have an internal index, but Full Text Search was much more efficient, especially since she could search by “Samuel Daniel” and “Royston.” So that was fun. How was that to discover all of those right away?

Diana (13m 18s):
Oh, it was great. It was exciting. And you know, we’ve talked before about what do you do if Full Text Search can’t quite read the handwriting. So there are some strategies you need to know about. And one of those is the naming. So I’m using Samuel Daniel and those could both be first names, right? So one of the tricks that I learned was put it in quotation marks. So it will look for “Samuel Daniel” first. It will still pull up some of the Samuel, Samuels or just Daniel hits later. But you will at least get the ones at the top that have those two names together.

Diana (13m 59s):
And I like to just start off with a really broad search within that microfilm or that IGN, image group numbers, collection. And then if you are starting to get way too many results, you can try to narrow it down. That collection was just order book 1772 to 1782. So it was already pretty specific, but sometimes I will start with just putting in, you know, just the name and then you maybe get thousands of results and you can start narrowing it down. That’s actually really helpful if you don’t know what county your person was in or a location. So you know, you, you need to figure out specifically what you’re searching for.

Diana (14m 42s):
And in this case I knew it was Middlesex County So I could go right to the microfilm I wanted to search and then also try variant spellings. And I tried just Samuel, I tried just Daniel, I tried just Royston, you know, I wanted to make sure I was getting everything And it did a really good job. This handwriting was clear, it was small but I did a good job reading it and finding all the different hits. And I did notice that internal index is unusual for court records. But I did go and check that and I don’t think that all of the records were actually indexed.

Diana (15m 22s):
I feel like I, there were only a few that were in the index and the rest of them were just scattered throughout the court sessions. So you know, we need to maybe check indexes as well as Full Text Search. We may have to just browse if we’re not finding what we think should be there. But it was so great to get 12 records, 12 different court records and you know, that’s the first part. But then you’ve got to actually make sense of them, get them transcribed and do something with them.

Nicole (15m 54s):
Alright, so what do we do with them? Well the first step is transcription and each court record entry was transcribed using Google AI Studio. And this is just one of our favorite tools. It has access to the best handwritten text recognition model, which is actually a multimodal large language model, Gemini, Gemini 3. Full Text Search provides a transcription that helps when browsing the record and finding it. But once you determine that it’s relevant and you want to actually use it as evidence, it’s good to then download the record to your files and then upload it to be transcribed somewhere like Google AI Studio, which is free.

Nicole (16m 34s):
And this provided the AI with the best copy possible and also saved the image to her own files. So it’s a good practice of course to download your images. And then if you’re gonna use AI, you need to have a high quality image, you don’t wanna take a screenshot, you wanna make sure, I mean you can take a screenshot if you’re zoomed in, but if you’re doing like a whole page, definitely download it. And she used a very simple prompt and she likes to provide the names and always add “preserve the line breaks so I can easily check for accuracy.” So the script was actually not easy to read. The letters were kind of small and close together. So she used the prompt, “Transcribe this court case between Richard Wyatt Royston and Samuel Daniel.

Nicole (17m 18s):
Preserve the line breaks,” and it’s easy to do, she already knows their names so it’s good to include those because then if they get the name wrong, they you’ve already told it so that it can know that context. So any context you can provide easily and simply to the AI tool, it really helps it. So Gemini now knows the names and so instead of transcribing you know, Richard Coyden or something wrong, it can know, oh yeah this is Royston. You can open the AI’s thoughts to see the process in Google AI Studio, which is fascinating. Some of the thoughts are, “beginning transcription; I’ve started transcribing the court document focusing on the case of Richard Wyatt Royston and Samuel Daniel;

Nicole (17m 59s):
I’m taking it line by line and noting the script; It’s a challenge, this old cursive; and it says, analyzing legal text; I’m now carefully dissecting the legal jargon; I’ve broken down the first few lines noting that PLT stands for plaintiff and DEFT for defendant; I’m focusing on the parties involved highlighting Royston and noting the court’s action regarding Richard Wyatt Royston since he’s an absentee; the transcription progresses even though the cursive is challenging.” So you’ve gotta love that the AI speaking as if it were a human about how challenging it is and noting some of the challenging abbreviations.

Diana (18m 38s):
Right. I thought that was really fun to see its thinking process and you don’t see that unless you open it up in Google AI studio. So it’s kind of hidden, but you can click the little down arrow and then you can see it. But it did a great job with the preserving the original structure. It was very readable and I was really glad because I didn’t have to read it and transcribe it myself. ’cause it was kind of difficult to read for me. But AI could see it perfectly.

Nicole (19m 7s):
Yeah, it’s great because it is challenging to be the, the first time you read through it to get used to the script. Once AI does that first step for you and then you’re checking for accuracy, you already have an idea of what it’s supposed to say based on what the AI guessed at. So then just kind of doing a quality check and you can clearly see in the transcription if something seems off or you just know I need to check all the names and dates and places because those are common error places. And so it really speeds up the transcription process, but we still always check for accuracy.

Diana (19m 40s):
Right. And I also noticed that some of the terms like court records can have some different terms in them that I’m not familiar with and I was kind of struggling to see what that word would be. And AI has that in its dataset, you know, the different legal terms and it no problem, you know, transcribed the correct legal terms, which was fabulous and helped me a lot since that’s some of the, that’s, you know, lacking in my knowledge is a lot of the old legal terms or even the the current legal terms. It’s just not language we use all the time.

Nicole (20m 14s):
Right, all the legalese and the abbreviations and just all of that stuff. There’s been so much of that on the internet with, you know, various things that the AI was trained on that it sure does have a lot of knowledge about legal terms. Alright, let’s have a word from our Sponsor. Are you ready to take your research beyond names and dates? If so, Newspapers.com is a game-changer for genealogists of all levels. With access to over a billion pages of historical newspapers, you can uncover rich, untold stories about your ancestors that you won’t find anywhere else. Imagine discovering a detailed obituary that reveals the extended family, or a local news story that highlights your great-grandmother’s role in the community. Maybe you’ll come across a photo capturing a moment in your ancestor’s life—something that brings their story vividly to life.

Nicole (21m 0s):
Newspapers captured the everyday details—announcements, happenings, achievements, and challenges—that paint a fuller picture of who your ancestors were and what their world was like. These primary sources reveal the social history behind the names on your family tree and help you connect with your past in a personal, meaningful way. Searching is simple and intuitive, letting you explore by name, location, or date to unearth those hidden gems. Ready to enrich your research with stories you won’t find in traditional records? Visit Newspapers.com/FamilyLocket today and get 20% off a subscription. Start turning your family tree into a family story.

Diana (21m 35s):
All right, so now let’s talk about logging the research and getting all of this information compiled. So I was using my Airtable research log and I wanted to enter each one of those court orders and there were 12. I arranged them chronologically and many of them were just in the same year. So to keep them straight, I wrote the dates as 1773-April, and then after AI transcribed the record, I added that to the research log as well. And then it was pretty easy to do the citations because they were all from the same image group on FamilySearch. So I could just duplicate the source citation that I had created the first time, and then I would adjust the page number, image number names, dates, and URL.

Diana (22m 19s):
So I had a system going and I had them all in my research log. And after I had all 12 court entries there, then I opened up a Google doc and I copied and pasted in all of the different transcriptions. So I had that little, I created a little title for each entry, such as April, 1773, court order Royston versus Daniel. And then I also put the search citation and the FamilySearch link and document dates and page numbers. So that document, having everything all together really was what I used for my deeper analysis because even though I had these records, I had to try to make sense of them.

Diana (23m 5s):
And it was challenging. There were just all sorts of different things going on. It didn’t seem like it was all the same case. So I uploaded that document, which was my compiled court record to Claude AI, and I asked it to do a chronological table showing the date, parties involved, the action and outcome for each entry. So even though I had this in my research log, I wanted to have a comprehensive table that at a glance I could see the basics of the case. And so it did a great job on that. You know, it was just using my specific information I’d already compiled. And so I was happy with, with the table.

Diana (23m 45s):
I feel like that’s something AI does really well. If you’ve given it the data and then ask it to do a specific table, it can do a nice job. So you know, what I found was it started in 1773 and then it went through 1774 and there were all sorts of different cases. There was a Richard Wyatt Royston, who’s John’s father, versus Samuel Daniel. And then there was a record of Samuel Daniel versus Richard Wyatt Royston and John Royston, which I would’ve expected to see. So we had a few different cases And it was confusing to me, but again, I was able to use AI to help, help me figure it out.

Nicole (24m 36s):
That’s great. I mean, without the help of AI, sometimes these things would just kind of sit in our research log like, okay, I transcribed all this, it’s kind of hard to understand it. What does it actually mean? Takes a little bit more work and effort. So it’s nice to have a little boost in that with AI. Well, while AI dramatically accelerates core records research, we can maintain professional standards by doing a little extra work, verifying everything we mentioned. Always check AI transcriptions against the original images, especially names, dates, and places. Then cross-reference AI explanations of legal terms with historical legal dictionaries.

Nicole (25m 20s):
So while AI has a lot of knowledge in its training data, if you’re still learning about legal terms and you’re not able to verify based on your past knowledge, it’s good to check the legal dictionaries and just to get familiar with those terms and start to get more expert at that. We don’t always wanna just rely on the AI’s explanations, it’s kind of a stepping stone to get us to where we need to go. And sometimes where we need to go is a legal dictionary to confirm and cross-reference with then confirming relationships and conclusions through traditional genealogical methods. Anytime AI is analyzing a source from the past and trying to extrapolate evidence about relationships, those are just guesses and oft, they’ll skip over the most obvious answer and give you some other ones that aren’t as good as far as like relationship possibilities.

Nicole (26m 9s):
So it’s good to just use other traditional methods as well. You don’t just only wanna rely on the AI. Document your process. I love this one because it’s not something I remember to do it, you know, all the time, but I’m trying to do better with it. But it’s to record which AI tools you’ve used and when and save important AI conversations or chat bot conversations as a part of your research documentation and include AI assistance in your methodology notes. And since we’ve been talking about Google AI Studio, I recently learned that free users of AI studio, only 100 of your recent chats are saved. So if there’s something that you want to keep track of, definitely copy and paste that into Google Doc, a Word document, export it as a PDF, something like that.

Nicole (26m 57s):
I’m usually copy and pasting everything into my research log. But just be aware that you don’t get unlimited chat history in Google AI Studio. Another thing to record as you document your AI process is to include AI assistance in your methodology notes. So just disclosing the fact that you’ve used AI and where, it can be so important. And then understand the limitations of AI, it can miss records. So manual review is important to do still. You can go through those court books and just make sure you haven’t missed anything. Especially if you feel like there’s a gap in the minutes. You know, as you put everything together and you, something seems to be missing can do that. Remember, the AI can often make incorrect assumptions.

Nicole (27m 40s):
The burden of proof still lies with the researcher. And finally, cite appropriately. Be sure to credit AI assistance in your research notes, even if it’s just for yourself, because you’ll want to remember where you did it. Sometimes I’ll think, oh, I used AI for this. Where did I do that? I wanna go see it. And then I can’t remember if it was Claude, Gemini or ChatGPT or Google AI Studio. So it’s good to make a note about that for yourself.

Diana (28m 3s):
Oh, absolutely. I do that all the time. I have to remember where I found something or where I chatted and it’s such a good note about Google AI Studio and those chats disappearing ’cause I did notice that myself. I was working on something and thought, I think I have already done this. Where is this? And it was gone. So another good reason for having good research notes.

Nicole (28m 24s):
Yeah, and I like your tip to link to saved conversations when possible. So in Claude, ChatGPT and Gemini, those that chat history shouldn’t disappear at any time, so you could save a link even in your research log to just, you could even add a column to Airtable that says AI chat bot conversation link.

Diana (28m 44s):
Right? Or even in, you know, sometimes I put it in, so I have added a column next to transcriptions, right that have abstracts. And so I will transcribe, then I’ll have Claude do an abstract and then I will sometimes put the link right there so I can get back to the original if I wanna continue asking questions about it or chatting about it.

Nicole (29m 6s):
Perfect.

Diana (29m 6s):
Yeah, all these little tricks. So you know, just something to keep in mind, if you’re just getting started on using AI with court records, we often talk about having the role, the goal, the text, the task, and the format. So it’s always good to just tell the AI the role that you’re an expert genealogist specializing in 18th century Virginia Court records. You know, let’s just get it in the right data set or get it in the right frame of mind. And then the goal was, I need to understand legal proceedings between Samuel Daniel and the Royston family. The text was my compiled court records document, or I could have pasted in the transcriptions, but I wanted it all to come together, so I did it in that document.

Diana (29m 50s):
And then the task, create a chronological table showing dates, parties, actions and outcomes, and then explain the legal terminology and analyze what these proceedings reveal about their relationship. And then the format, present findings in a table, followed by narrative analysis include citations to specific court entries. So, you know, that might be seem kind of detailed. I have found that sometimes I can just be really simple with my prompts and sometimes the AI will do great with that, but other times, you know, kind of makes up whatever it wants to do if it’s too simple. So you know, if you’re not getting the good results you want, you might consider working on your prompt a bit.

Diana (30m 31s):
So looking forward, it’s so exciting to have these court records now available to us. They’re so amazing. And you know, in the past they’ve just been really hard to access and interpret. And so AI has really reduced the barriers to using them. And as FamilySearch continues, adding more collections to Full Text Search and our AI tools become more sophisticated, I think we’re going to see even greater possibilities. It was really fun to use the Samuel Daniel research, and I did it just in two weeks. It was so helpful to have AI’s help as my assistant in doing all that transcribing and analysis. So the key really is thinking of AI as your powerful assistant, but not a replacement for traditional genealogical methods.

Diana (31m 20s):
So, you know, use it to find records faster, understand them more deeply and organize them more effectively. But always, always, always verify and document and use your good principles of sound genealogical research. So you’re probably all wondering what did I learn from all of that? Well, we are gonna do that in the next episode where we will go through the court records and see how AI helped make sense of this case.

Nicole (31m 48s):
Alright, I’m looking forward to part two, that will be fun. Thanks everyone for listening and we hope you have a great week. We’ll talk to you again next week, bye-bye.

Diana (31m 58s):
Bye-bye.

Nicole (31m 57s):
Thank you for listening. We hope that something you heard today will help you make progress in your research. If you want to learn more, purchase our books, Research Like a Pro and Research Like a Pro with DNA on Amazon.com and other booksellers. You can also register for our online courses or study groups of the same names. Learn more at FamilyLocket.com/services. To share your progress and ask questions, join our private Facebook group by sending us your book receipt or joining our courses to get updates in your email inbox each Monday, subscribe to our newsletter at FamilyLocket.com/newsletter. Please subscribe, rate and review our podcast. We read each review and are so thankful for them. We hope you’ll start now to Research Like a Pro.