Thrifty Token Tree Tips

Now that I have become a Claude convert, I decided to learn about the models and usage so I don't end up torching cash like Amazon did. Granted, I would run out of cash much faster than Amazon, but still. Hitting usage limits is super annoying! Found this GREAT article with all kinds of tips like "Chat uses less tokens than Co-work" and when you are working on something, Claude will re-read everything all over again which uses more tokens so you can save tokens on a Co-work session just by using this prompt suggested by Ruben Hassid:

"At the end of a Cowork session, prompt “Write a session-notes.md with the key decisions and next steps.” Next session, start with “Read session-notes.md first.”

As so often happens with technology, some of the tips Ruben provided four months ago in April 2026 are already out-of-date due to changes made by Claude. Like, I tried to change the setting to "concise" but it's gone. Yes, I know it partly moved to "Customize->Skills" but "Concise" would need to be manually added.The next burning million-dollar, er...I mean million-token question is which model do I use for what? Ruben gives this advice:

"Grammar checks, brainstorming, reformatting, short answers. Sonnet handles all of this at a fraction of the cost.

Opus + Extended thinking is your heavy machinery. Don’t use heavy machinery to move a chair.

My rule: if the task takes Claude less than 30 seconds to answer, it probably doesn’t need Opus. Switch models before you start the session. It takes 2 clicks."

Remember, his post was written before Fable launched so put Fable under the "complex task" and "token-torching" category in your grey matter.

And speaking of token-torching, I tried out Ruben's tips for NOT uploading images to Claude and torching my tokens that way - I have figured out a way that I could obtain the info I needed for Claude -AND- save tokens in the process. The two apps I used for this were FamilySearch and

the "Notes" app that comes standard in my iPhone. I regularly use the Notes app for voice transcription because it's pretty good. To do this test run, I first opened my FamilySearch tree to Adam Weaver where I know I have saved several pages of a Weaver book in the sources section of Adam's profile. The book is almost 100 pages and as I recently hit that 100 page limit in another chat, I have been thinking about capturing the information in this book without hitting that limit. So here's how I accomplished that and the first test run did not even change my usage percentages! Woo-hoo! Open FamilySearch on your phone, find the image of a source you want to save, click the "copy" icon on the transcription (circled in red in the left screenshot), then paste that into a new note on your phone. Caveat, I was copying and pasting a typed transcription so FamilySearch's transcription had little to no errors. If you are copying transcribed handwriting, there will likely be a higher error rate. I copy and pasted three transcriptions into the Note and then clicked "Select All" in the Note and then pasted that into Claude. I included the prompt in this image and it put the pasted text above my prompt but it still worked like a charm. 

I also have a TON of books and other genealogical resources I have imaged with my phone. So I tried an experiment and searched for "Adam" in my photos, found a page from the same Weaver book, held my finger down on it and pressed down until "Select all" came up, Copy, and then
pasted into the Sonnet 5 chat. It updated with the additional information just fine but as Ruben

points out, it uses more tokens when Claude has to keep re-reading through previous chats. It does also work to just copy and paste the text into the iPhone note, which to reiterate, saves usage over just constantly updating the chat.

Anyhow, between Ruben's tips (I recommend reading his whole post) and my tweaks for genealogy, I hope this saves you some tokens. Don't forget to proof the work for accuracy.

Are we really looking into the eyes of that Pompeii victim?

In today's news, the Pompeii Archaeological Park in collaboration with the University of Padua announced an AI "reconstruction" of a Pompeii victim from Mount Vesuvius' AD 79 eruption. While artificial intelligence can be a more cost-effective and quicker way to do facial reconstructions than clay models, it is hoped that prompters will do every possible measure to insure their interpretation from remains is not altered or improved upon by an AI. Processes and tools should be peer-reviewed and published in academic journals. If you find that this particular case has indeed been published in a peer-reviewed journal, please share the link in the comments section below. While genealogists have photos that can be regenerated, photos produced with AI should always be disclosed and labeled. See the Coalition for Responsible AI in Genealogy's photo statement for more information.

To edit or not to edit? Disclosure is the answer

Salt Palace 2026

In today's news, the prompter and motion capture provider of AI creation, Tilly Norwood, is receiving death threats. An interesting point the prompter makes is that "Tilly is famous. I'm not. That's wonderful. Fame is a horrible thing." Death threats definitely add credence to the "fame is a horrible thing" posit. Norwood's prompter also makes the point that since the AI is the famous one, she doesn't have to do things like Botox to preserve youthfulness. Altering one's appearance was a recent discussion I had with Ilya at The Family Vault (an AI Agent and cloud storage) booth at 2026 RootsTech. Ilya questioned why disclose using AI for photo regeneration when people edit photos with Photoshop or even that many women wear make-up and that's editing their looks? I've been mulling this over ever since RootsTech! In the Genealogy Community, it is actually an asset to be older. When I became active in genealogy at age 31, I actually had genealogists be dismissive towards me due to my youth. Some genealogists wear make-up and some don't; it's not a field where appearance is judged like it is with celebrities. I have noticed that make-up tends to help when on Zoom meetings or on stage because the bright lights tend to wash out facial features. Unlike with AI use, there's no need to disclose I'm wearing make-up because it's quite obvious and visible. Of course I didn't think of this at the time I was chatting with Ilya - the case for disclosure lies with whether editing can be detected or not. Make-up is visible. Tinting old photos is visible. AI photo regeneration is not visible (at least to the naked eye). Read more here on the Coalition for Responsible AI in Genealogy's "Protecting Trust in Historical Images" position statement.

The Gamechanger: Using FamilySearch Full Text AI

My FamilySearch Stats

Several weeks ago, my husband asked me what I was working on. I replied that I was working on my genealogy and his response was, "Why? I thought you were done." Now to a genealogist's ears, that's a funny remark because all genealogists know that we will never be completely "done" as there is always something to work on in the family tree. Granted, I did the vast majority of my computer tree research in the early 2000's and then pivoted to DNA research. The only time since that I have conducted traditional genealogy research was when I had the opportunity to do in-person research at archives like the Daughters of the American Revolution Library, the Clayton Library in Houston, and the South Carolina Department of Archives and History, etc. etc. But the reason I now have a laptop in front of me in the quiet hours of early morning and late at night is thanks to FamilySearch.org implementing Full Text AI Search. In 2025 alone, I broke through an Irish ancestor's brickwall that we had thought would never be broken considering the difficulty imposed on research by the Four Courts fire in 1922 and I found another ancestor during his "lost years" in another county. In my mind, FamilySearch Full Text AI is the best thing since sliced bread! So if you're into family tree research, and you have not tried it, you need to and I will tell you how I get good results. But first, I highly recommend you have already connected to a tree in FamilySearch. For most people just starting out, that usually just entails connecting yourself to your parents and sometimes grandparents and the tree is usually already built out. And if you are not familiar with FamilySearch, it is entirely FREE! As you can see on my stats graph on the right, the huge spike in 2025-26 contributions (which is mostly records I have attached) is thanks to Full Text AI. Connecting yourself in a tree makes for ease of navigation, and will connect you to cousins at "Relatives at RootsTech."

To start your first Full Text AI search, log in to FamilySearch.org - then go to the top menu and click "Search" on the upper left menu, then click on "Full Text". When the box comes up, enter your ancestor's name in quotes, e.g.: "Rody Kennedy" then the place they lived "South Carolina" and a year range (I usually make mine a little broader than their birth year and death date). Then submit. Sometimes, especially with a common name, I will add another name related to that ancestor in the "Keyword" box above their name, for example, his wife "Ellen" and that will narrow the results down. Putting the name in quotes works much better than using a plus sign as I have tried both but I learned about the quotes from this video on FamilySearch AI which I highly recommend. This video is not only excellent for learning about this tool, but Colette Hokanson includes great general AI info as well. After you found a document that matches your ancestor, you can download a JPG of it and a PDF with or without a transcription. Be sure to proofread the transcription and correct errors before attaching to your ancestor's profile. I leave or add the words "Generated from AI summary" or "Generated from AI transcription" for disclosure purposes. Sometimes, genealogists like myself have already attached the record and it will show you if it has been attached, but do click and verify who the record was attached to. I recently attached a will of a friend of my ancestor who had named her in his will. You can also correct spelling errors by clicking the little pencil icon at the top right corner of the transcription box.

I have to say, I am deeply grateful to FamilySearch for implementing this AI Search function as it has allowed me to find records that I never would have found in my lifetime otherwise. On my very first trip to the FamilySearch Library in Salt Lake City in May 2023, I wasted an entire day of my life there looking for my Bohemian Great-Grandfather, August Bauman (wife is Josefa Engel) in the microfilms. When I could not find these two needles-in-a-haystack, I resigned myself to waiting for human volunteers to index these records for a kingdom that no longer exists (Bohemia that is). While these European collections have not yet been added to the AI searchable collections, I now have hope they will be and I will find them.

RootsTech 2026 is almost here and one of the exciting things about RootsTech is all the company announcements and new products/features. You can attend RootsTech online for FREE - sign up here and be sure to add lots of great AI sessions to your watch list including the session on "Guidelines for the Responsible Use of Artificial Intelligence (AI) in Genealogy in 2026".

Free or not to free

Have you ever heard the saying, "If something is free, you're the product"? You innately know that's true. That's why Facebook has ads and listens to your conversations (which is a super-annoying invasion of privacy I might add). And this is why you can pay for ad-free YouTube. Well things are no different with "free" Artificial Intelligence (AI) apps. Instead of shoving ads in your face, you are the bug in their jar. What you input into AI is used to train the AI - you become the training data. It's the price you pay for using their products for "free". However,  even some of the subscription based apps still use your data for training but at least you might be able to opt out of it like with ChatGPT. Same with the paid versions of Google Gemini. So if you are using AI to process personal data, like family trees, DNA results, etc. then be sure to remove all data on living persons so it doesn't end up in the training data. The Coalition for the Responsible AI in Genealogy (CRAIGEN.org) provides a good principle to follow for this: "AI usage can lead to unintended data exposure, putting private information at risk of being publicly disclosed. Therefore, members of the genealogical community take reasonable measures to safeguard private information when using AI." One more thing to mention on the topic of training data: if you are following this blog, I wrote yesterday about how the data that AI produces is from a compendium of all human knowledge. Most of that human knowledge is gained from the AI companies sending out bots to scrape the information from websites. While these would seem harmless if the data is publicly available (unlike Perplexity sending their bots behind paywalls), it does create a huge load on a site's servers. This is why you are seeing such an increase in "Verifying You're Human" captchas. The scraping bots were overloading our server on the International Society of Genetic Genealogy (ISOGG) so we had to implement one of those captchas. The captchas are annoying (but not as annoying as Facebook spying) but at least you now know why we have seen such an increase in use. So go easy on sites that implement those, we should not have to pay a server more for information we provide for free just so an AI company can turn around and charge for it as data they provide!

By the way, I gave ChatGPT 5.2 free license (very little prompting) to create this graphic - pretty good I'd say!

AI and Genealogy (not quick, nor dirty) Glossary

There appears to be multiple options for when Artificial Intelligence was created; possibly by Alan Turing or at a conference in 1956. Either way, it was developed mid-20th century. Although companies (including genealogy companies) had been using AI for many years, the official launch date for use by the general public is November 30, 2022 when OpenAI debuted ChatGPT. Since that time, a whole new AI lexicon has been trickling forth; and in some cases, the trickle is more like a flood. I will be posting the new terms as I learn them, but for now, I had ChatGPT 5.2 Thinking version create this handy graphic. It did take me five tries to get the graphic where it's at. The first two attempts had wasted space, and when I prompted it to fill the space, it repeated the terms to fill it. (Facepalm) The last attempt also had wasted space so I asked it to move the citation/disclaimer up under the boxes. Interestingly, the first three attempts had colorful graphics and used several other AI terms. I figured I was dealing with "context rot" so started over and it gave me the plain vanilla result you are seeing. I'd rather have accurate than flashy so it is what it is. I did have ChatGPT provide sources for the terms. Accuracy, accuracy, accuracy - ALWAYS proof AI's work or you might end up in an unfortunate news headline.

 

Adding Custom Instructions to ChatGPT

This morning, I was discussing AI with one of my longtime genetic genealogy friends and he replied that he recently subscribed to ChatGPT. I asked him if he had added "Custom Instructions" yet and he said he had not; so I replied that I would do a blog post about it as this is something I highly recommend everyone does. To customize your instructions, open ChatGPT, then click on your name, then click "Personalization" and under the box that says, "Custom instructions" and enter the following or you can edit it to tailor it to your needs. Here is what mine currently has: 

Scholarly, factual, no embellishments, no fabrications, no dashes, always cite sources. Do not search nor include any sources from Facebook, Instagram, X (formerly Twitter), and Grok. They are to be excluded.

I specifically exclude these sites because ChatGPT does give me these results if I do not. It still sometimes gives Facebook links and then I will tell it not to but it is much rarer with the customization. The hallucinations/fabrications are almost nil now. I still check all of the sources before I use them, and the quality of the sources has improved immensely. Definitely follow the Coalition for the Responsible AI in Genealogy Accuracy principle: "AI can generate false, biased, or incorrect content. Therefore, members of the genealogical community verify the accuracy of the information with other records and acknowledge credible sources of content generated by AI."

To give credit where credit is due, I learned about the customization from Lissa DuVall last December 2025 in the Genealogy and Artificial Intelligence Facebook Group so big debt of gratitude to her!