Hello friends,

Welcome back. If you missed last week’s edition, we announced some big news: AI for the Rest of Us has evolved from just a newsletter into a full business. We’re now helping executives and companies figure out what to pursue and what to ignore with AI. Check out the new website if you haven’t already, and if you know someone who could benefit from what we do, please introduce us. I’m always around at [email protected].

Back to the newsletter.

You might recall that we’re in the middle of a six-part series based on a LinkedIn post I wrote back in December with 22 pieces of advice for getting more out of AI chatbots. Each week, we’re unpacking a few of those ideas in more depth.

Last week (week one) we talked about the mindset shift you gotta have working with this chatbot, mainly that they’re not a better Google, they’re more like thinking partners. This week, we’re going deeper on that idea. Because if they’re thinking partners, that means we should consider... talking to them.

Yup, talking. Not typing. Talking out loud, with our voices. And not just talking – arguing, pushing back, asking for a POV we don’t wanna hear, etc. I’ll even share what happened when I asked a chatbot to roast me based on several months of meeting notes. (I’m still recovering in case you’re wondering.)

Here we go.

– Kyser

Confession time: I’m not writing this newsletter. Well, I’m not not writing this newsletter. Just this part of the newsletter. I’m speaking it out loud, into my phone (using an app called Wispr Flow), while getting some air in the middle of my work day. I have not altered the formatting or any of its grammar, not even those parentheses in that last sentence (I did come back and add the hyperlink though). It does it automatically using the big bad AI technology. It even knows when I should have a list instead of just a sentence. For example, I need to get a few things from the grocery store later today:

It made that list for me as I talked. You get the idea. Let’s go back to using the fingers, shall we?

One of the best ways to use AI – period – is to use an AI voice tool. I’m not talking about using a chatbot with their voice mode features (we’ll get there). I’m talking about an app that’s downloaded on your computer and built into your phone’s keyboard. You use it instead of your fingers on a keyboard and your thumbs on a screen. (We all know our thumbs need a break after the past 15 years!)

You use these programs to talk out your thoughts instead of writing them. To dump everything in your brain into words without your (tired) thumbs getting in the way. To get your thoughts outta that noggin of yours.

Not convinced? Keep reading.

Did you know you speak way faster than you type? I kinda knew that – either intuitively or someone told me. A Stanford study from a while back found that it’s about 3x faster. We speak at around 150 words per minute. We type at around 40. That’s a huge difference – and after using some of these tools, I get it.

But speed isn’t the main thing. When you talk, you think differently. You don’t edit yourself as much. You ramble a little. You go on tangents. You say, “actually, wait, here’s what I really mean.” You get the messy, unfiltered version of what’s in your head, which, it turns out, is exactly what you need sometimes.

When I type, I tend to compress. I overthink and I under-think at the same time. I leave out context because typing can sometimes feel like work or like I need to get it exactly right. But when I talk? I over-explain. I give background. I unload all kinds of thoughts. And it’s actually very freeing.

Apps like Wispr Flow and Monologue are insanely good at this. You talk, they transcribe – and it’s fast and accurate. I use Wispr Flow all day every day to brain dump. I’ll go on for two or three minutes straight, getting all my thoughts out, and suddenly I have a wall of text that actually captures what I was trying to say. Oh and bonus: the recipient isn’t sitting there yawning or annoyed. At least I don’t think.

Now here’s where the chatbots come in. That wall of text? Paste it into a chatbot. All that context you just dumped out of your brain is exactly what the AI needs to give you something useful back. These chatbots have gotten really good at making sense of messy, rambling input. I like to think one of Claude’s main jobs is to organize my thoughts. And it’s very good at it.

With chatbots, you can also skip the middleman entirely (a la an app like Wispr Flow). Most of the major chatbots have voice mode now — ChatGPT, Claude, Gemini. You tap a button and have a spoken conversation. The AI talks back. I’ve mentioned this before, but it’s still one of the most underrated features of these tools. Most people don’t even know it exists.

Try it on your commute. Try it while you’re cooking dinner. Try it when you’re stuck on something and don’t even know how to phrase the question yet. Whether you’re brain dumping into Wispr and pasting into a chatbot, or going back-and-forth in voice mode directly, the point is the same: when you need to think out loud, it helps to have something that can think back.

Now, before we go on, and before you start sending me hate mail: I’m not advocating for voice to replace good ole fashioned writing. That irony is not lost on me. I’m writing a newsletter after all. I love writing, and still believe it’s critical for our brains and all around psyche. It’s important and healthy. I insist my older child writes in a journal. I’ll never stop writing. I could go on and on, but how about I just continue writing a newsletter? 😉

Speaking of speaking to AI, here’s something people don’t do enough: argue with the chatbot.

I don’t mean yell at it. (Though I do that sometimes. Please don’t tell my kids.) I mean push back. Disagree. Say things like, “I’m not sure that’s right” or “that doesn’t match my experience” or “what about this other angle you’re not considering?”

These tools are trained to be helpful, which sometimes means they’re trained to agree with you. They’ll validate your ideas. They’ll tell you your plan sounds great. They’ll be very polite about the whole thing.

That’s nice, but it’s not always useful.

The magic happens when you treat it like a colleague you respect but aren’t afraid to challenge. When you say, “I don’t think that’s the best approach — here’s why” and watch it reconsider. When you say, “You’re being too generic, give me something more specific” and it actually does. Or my favorite, “You sound like an AI bot, please stop.”

The chatbot isn’t offended (let’s hope 😬). It doesn’t have feelings to hurt. It just adjusts. And often, the adjusted answer is significantly better than the first one.

I do this constantly. I’ll get a response, read it, and say “that’s fine but it’s missing the emotional component” or “you’re overcomplicating this” or “I disagree with your second point.” And every time, it comes back with something more useful.

Another way to think about it: the first answer is your starting point. The real value comes from the back-and-forth you do with it.

This might be one of my top 3 most and least favorite prompts: “Tell me what I don’t want to hear.” Favorite because it’s so good for me. Least favorite because I legit don’t like the responses. But that’s the point.

It works because these tools, by default, are agreeable as we said above. They’re designed to be helpful, and helpful often gets interpreted as supportive. You share an idea, they find the good in it. You share a plan, they help you execute it. You share a rant, they validate your feelings.

That’s fine sometimes. But sometimes you need the opposite. You need the friend who says, “Come on, what are you doing??” or the advisor who says, “Have you considered that this might not work?” Or the therapist who sits across from you in a room, breathing the same air you’re breathing, sensing your emotions, sometimes even feeling them, and asking you to explain more. And by the way, nothing can replace those things. Ever. Not a chatbot. Not some companion app that’s invented next year. Not even when we get the real-world version of Kitt or Samantha. I’m talking about “lower stakes” things here.

For example, I’ll share a business idea and say, “now tell me what I don’t want to hear.” I’ll draft an email and say, “what’s wrong with this that I’m not seeing?” I’ll describe a decision I’m about to make and say “play devil’s advocate — why is this a bad idea?”

And it delivers. Not because it’s trying to be negative, but because you’ve given it permission to offer a perspective it wouldn’t give by default. This is an important thing to know about these tools. None of us should fall prey to their sycophancy (fancy word for being a yes-man).

There’s one more incredibly important topic to touch on here. Your own personal roasting machine. 😊

These things are eerily good at it. Like really good. As in, “I feel personally attacked but also I’m (mostly) laughing” good.

I’ve got a fun example for you. I’ve met with over 100 business leaders and executives over the past two years, sometimes in formal conversations where I’m talking about who I am, what AI for the Rest of Us does, how we think about AI, etc. I’m usually recording these meetings with a fancy AI notetaker like Granola or Otter (with permission of course).

Well, I decided to upload several months’ worth of meeting notes into a chatbot and ask it to roast me based on the patterns it saw. I can’t believe I’m sharing this, but yolo.

Don’t you dare share that with my wife or kids. It’s already affecting my soul a bit. And yes, my parents named me after my middle name, it’s a thing, let’s keep moving.

In all seriousness (mostly), here’s why this matters beyond just entertainment: it’s a genuinely useful way to see yourself from the outside. We all have blind spots. We all have patterns we can’t see because we’re too close to them. And while a roast is exaggerated for comedic effect, there’s usually a kernel of truth buried in there. In my case, an XL-bucket-of-buttered-popcorn-at-the-movie-theater amount of truth.

So yeah, let the chatbot roast you. It’s fun. It’s humbling. And if you’re brave enough to sit with the discomfort, it might actually teach you something about yourself. If you’re even braver, send me your roast to include in next week’s newsletter 😜.

To recap: Talk to these things out loud. Push back when they’re wrong or generic. Ask them to tell you what you don’t want to hear. And occasionally, let them roast you for sport – and good learning.

These are all variations on the same theme: treat the chatbot like a conversation partner, not a search bar. The more you engage with it, and I mean really engage with your voice and your opinions and your willingness to be challenged, the more useful it becomes.

And that wraps up another edition of AI for the Rest of Us.

Next week, we’re flipping the script slightly. We’ve been talking about how to get more from AI. But what happens when AI starts doing too much? Specifically, when it starts sounding like you – and you didn’t ask it to.

We’ll talk about why copy-paste is almost always a mistake, how to spot AI slop (including the tells that give it away), and how to keep your voice yours in a world that’s increasingly automated. Fun stuff!

Until then…