Issue #2 [2026 - Week 1]
Welcome to the second issue of 3 Things AI!
Like a lot of you I’ve been using ChatGPT or similar tools for various questions and tasks and inquiries. I figured I’d start capturing three things I did that were new or interesting each week using AI tools and share them. If you enjoy reading, subscribe to the newsletter! And please send me your favorites! Once a month, I’ll pick reader suggestions and share out a bonus newsletter (thanks for the ideas you’ve already sent!).
I was very excited about item #3, so that one is a bit long, but I’m sure readers will find it useful and fascinating as well. I have also linked to a number of sites or tools - just to be clear I have no affiliations with any companies or products I am just sharing the stuff I find.
I used ChatGPT to prep for calls - at home and at work
Last week brought two powerful examples of how AI tools can help arm you with the right information and approach - even if you’re in a pinch. I used this capability for an IT/tech support call at home and a sales call at work.
Calling Apple support at home: We updated the operating system (OS) on our home Mac laptop this week, and immediately my son lost a bunch of important files from school (arrggh). I first prompted GPT with the issue and iterated through a bunch of troubleshooting tips, but nothing solved it. Finally GPT said that Apple Support was required, and it then gave me the instructions to contact them (Apple doesn’t make it that easy to find) and explicit language to ensure the support ticket was escalated. This worked! I moved through the automated system quickly and got an agent and I did not have to repeat all the same troubleshooting stuff. For example, how many people would say this: ”I need Tier 2 support to check server-side snapshots and restore my Documents container”? My issue got escalated and Apple scheduled a day/time for a support call, where we fixed the issue!
Last-minute sales call at work: I work with electric utilities and commercial buildings (offices, schools, etc.) to help reduce stress on the electric grid, and right before we checked out for the holiday break, a facility manager (Jim) for a college campus reached out to schedule a meeting about participating in their local utility program. The call would be with Jim, my colleague who heads up sales, and myself (I run the programs), and we only had one hour to prepare. We had sparse notes from one quick call with Jim over two months ago, and we had all the eligibility information about the program.
My colleague pasted/uploaded it all into our company’s CoPilot AI tool (it uses ChatGPT’s models, but just keeps our data/info secure and private. So this could be done with ChatGPT as well.). She explained the purpose of the call, asked for an agenda, talking points, a high-level script for the call, and likely questions that Jim would ask and the associated answers. My colleague also asked CoPilot for a summary of any known energy or sustainability goals or announcements the college had made in the last few years.
Amazing results! We had a simple, organized, professional agenda ready to go, we had all the college’s specific initiatives summarized so we could speak directly to their goals, we had eligibility criteria mapped to their buildings, and then we guided Jim through the key talking points. In the end Jim was going to get approval to enroll in the program.
Now, could we have done this without AI? Of course, professionals in all fields have been doing sales like this for years. But the ability to have so much information organized, summarized and mapped to our sole purpose for that call, especially given the tight timeline, was incredible.
I used a powerful new tool for research
Do you ever want to keep track of a specific news topic, or certain events or happenings in your town or a favorite city? Or do you run a business and you want the latest market research on your space, your customers, or your competition? Well, new AI tools are making this much easier.
A friend showed me Yutori, which is a free service* that keeps an eye on parts of the web for you, pulls out relevant details, and summarizes it in plain language so you don’t have to constantly check sites or read everything yourself. The possibilities are endless here. And the key thing is you can use your own natural language to type in what you want to search for and keep track of. It summarizes that and then creates a “scout” for you - that gets automatically updated when new things pop up and emailed to you. You can adjust the search terms anytime as well as the frequency for how often you get updates.
I immediately set up some very specific market research scouts for the field I work in (utilities, buildings, grid, etc.) and the results are impressive. For fun, I also asked it for lists and news on small and medium businesses in Burlington and Chittenden County, Vermont. You can see that newsletter for yourself here. Look at their homepage for example scouts, and try it out and see what you think!
*Free includes one “scout” at a time but you do have to pay for a monthly plan to have more scouts.
Cataloging
This is a huge topic that I stumbled upon this week, and one that can apply to everyday things at home or even form the seed of a new business. Take this story for example: a Mom of four young kids bought a pallet of hundreds of books for $1 so she could get books for her kids. She soon discovered that she could re-sell the curriculum books (the ones she had no use for) on eBay, and ultimately made $5,000 from that $1 purchase! The fascinating part - aside from the unexpected business she started - is that it was the existence of new AI tools like ChatGPT that made this possible. You can read more here and/or listen to the podcast (linked in the article), but she was able to take pictures of the books on the pallet and get full descriptions (e.g., author, title, publisher, year, edition, etc.) from ChatGPT so she could itemize her inventory and then price, list and sell them. This sums it up: “The reason most people don't do this is the sorting. That’s the moat. You’ll get hundreds of random books and have no idea what goes together, what's valuable, or how to price it. Before ChatGPT, this would take weeks of manual research. Now it can take minutes.“
So think about how many things you could apply this to. Do you have an attic full of books? Crates of vinyl records? Or what if you walked into a thrift store or a used bookstore, took a picture of the shelves in front of you, and asked for a list of what’s there? It’s not going to be 100% right, but it’s way faster than anything previously possible.
I decided to try this with some vinyl records I own. I’ve been nagging my kids to help me catalog these for years in a spreadsheet, but we’ve only ever gotten through one crate. It’s tedious and we’ve always procrastinated finishing the job. So I pulled 13 random records off the shelf and spread them out like a deck of cards, snapped a photo, uploaded to GPT (using the app on my phone) and asked it to “provide as much info as you can.“

First attempt: 13 records spread out like a deck of cards
First results: not bad but not great. 9 identified but the AI thought there were “~14-16.” It said to show the back covers or the spines. So I flipped them over and spread them out, and now asked for the results in a table form so I could put into a spreadsheet (GPT can also just create a spreadsheet for you, but I wanted to see the table). I also instructed it to not hallucinate (i.e., don’t make anything up if you’re uncertain), label “tbd” for anything unknown, and asked for label and catalog number and all the juicy details.

Second attempt: All records spread out and showing their back covers
Second results: Much better. Great table (sample row below). It still thought there were 15 (AI can be shockingly bad at simple math sometimes) but it identified at least the artist for 11 records this time.

Sample row from second attempt - showing back covers only
So I decided to try the spines only view. I didn’t even ask anything - just uploaded the photo. Now the results were cooking! I still only had 11 rows, as it apparently just could not read two of the spines. But the info on the the 11 was very comprehensive and 95% accurate (It called the Travelling Wilburys the “Wilburns” despite correctly naming them in another cell). You can see the table it created below. I can now catalog my collection in a fraction of the time. If I was trying to sell, this would make it super easy. If these were in a store and I was interested in buying, I would have a wealth of valuable information. And the bonus was that GPT complimented me multiple times, one time saying “This is already a very strong collection, especially the African & Ethiopian selections.” Hey, I’ll take it!

Third attempt: Photo of the same records but spines only

I bet readers will have many ideas how to use this capability - share them and I’ll include next time!
Well that’s a wrap. Send me feedback or comments and your own AI tricks and tips and hacks. There are a lot of tools out there I only use a limited few, so share what you’re using! If you’re curious how I used any of these, reply and I’ll explain. Stay tuned for next week’s 3 Things AI!
