[2026 - Week 5]
Welcome to Issue #6 of 3 Things AI - a weekly note where I share three practical or interesting ways I used AI this week.
Last week’s issue - focused on different AI tools I’ve been using - generated a lot of great feedback and examples from readers about interesting things they’re doing with these powerful new tools.
One of the dominant themes I’m seeing, whether people are using it for simple, everyday needs or complex work-related tasks, is that AI can always meet you where you’re at. In Issue #4, I wrote: “This is where AI actually allows us to be ourselves, to talk like a human and get the info we need at the time we need it.”
The examples below all demonstrate this - even across widely different use cases. A CEO organizing five years of professional relationships, a mom softening her texts, a college student learning physics the way that actually works for him. Different people, different problems, same underlying capability: the tool adapts to you, not the other way around.
Are there larger, existential questions surrounding our adoption of these tools? Yes, for sure, and these are widely discussed in the media. But for this newsletter, I'm focused on what's actually working for real people right now.
Building a personal CRM that actually updates itself
McGee runs a climate tech company, which means his network is everything: investors, partners, potential customers, advisors. But his contacts were scattered across calendar invites, thousands of emails, Fireflies transcripts (Fireflies is an automated notetaker for virtual meetings), LinkedIn profiles that may or may not match the person .
So he asked AI* to build him a personal customer relationship management (CRM) from scratch. If the term CRM doesn’t mean anything to you, popular CRMs include Salesforce, HubSpot, etc., which all require paid subscriptions and aren’t designed to take in complex inputs like past calendar invites and emails.
The ask: pull every calendar event and email from the past five years, match them with Fireflies transcripts, find LinkedIn profiles for each contact, categorize each person (investor? partner? customer?), count interactions, write relationship summaries, highlight and clean up the messy data (e.g., is "Bob S." and "Robert Smith" the same person?), dump it into Notion (Notion is a project management and note-taking tool and database), and write a script that auto-updates weekly based on new calendar invites and emails.
Now McGee has a living database of his actual network. He can see who he hasn't talked to in six months, which investors he's met with a dozen times, what his last conversation was about. And it updates itself, so it doesn't go stale.
This is really incredible that this can be done without paying huge fees for enterprise software or hiring a developer to build a custom tool.
*At press time I do not know which AI tool he used to do this. I’m guessing Gemini given the Google calendar + Google email integration, but I’ll report back!
Softening the delivery (without losing the message)
Dee is a mom with three daughters, and a self-described “direct communicator.” Which works great in some contexts and less great when communicating with her kids.
She's been using ChatGPT as a translation layer between her natural communication style and what actually lands well with her daughters. She'll write what she wants to say (direct, to the point), then ask AI to soften it before hitting send.
Her example: "Please clean the kitchen extra clean, your dad is coming home tonight."
ChatGPT’s version: "Hey sweetie, can you give the kitchen a really good clean tonight? Your dad's coming home from a long trip, and he loves it when it is an extra fresh space."
Same ask, totally different tone. One sounds like a directive, the other sounds like you're on the same team.
How many family tensions could be avoided if we all had a "soften this" button before we hit send?
Learning physics like you have a personal tutor
Kody is a college freshman and civil engineering major, taking physics and math. Instead of using AI to get answers, he's using it as a tutor that actually teaches.
He's specific about how he set it up. He told ChatGPT how he learns best, asked it to act like a distinguished professor, and gave it a critical instruction: don't just answer, explain at a fundamental level. He also told it which textbooks he's using (well-known ones that the AI already knows), so it can reference specific chapters and moderate its explanations based on what the book covers and in what order.
The big shift for Kody was asking "why does this matter?" and "what's the relevancy here?" for everything. Physics formulas or calculus techniques can feel abstract when you're just memorizing them. But when you understand why they exist and where they show up in the real world, they stick.
What he's built is essentially a one-on-one tutor available 24/7 that knows his textbook, knows how he learns, knows exactly where he’s at and won't just hand him the answer. It explains, then asks him to work through it.
Unfortunately, too many students are using these tools just to get answers, and they’re really missing a unique opportunity to enhance their learning. This is what AI should be doing in education. Not replacing the work, but making the learning actually make sense.
__________
If you have feedback, questions, or an AI tool you’re enjoying, just reply. I’m always curious what others are using.
I send this once a week. If it’s useful, you can subscribe here → https://3thingsai.beehiiv.com/subscribe