Stop the Slop: Five AI Fundamentals, Smarter Prompts, Real Results | Turn the Lens Ep45

Stop the Slop: Five AI Fundamentals, Smarter Prompts, Real Results | Turn the Lens with Jeff Frick Ep 45

English Transcript
© Copyright 2026 Menlo Creek Media, LLC, All Rights Reserved

‍Jeff Frick:
Hey, welcome back everybody. Jeff Frick coming to you from another forest in the neighborhood.

What I wanted to share was five tips that I learned in my AI class with KMo, Kyle “KMo” Moschetto, that I wanted to share with you. This is with his permission. He’s all about everybody getting smarter, and this is certainly something that he was happy to share.

Garbage in, garbage out. It’s the same that it’s been in computing forever. You’re only going to get out of it what you put into it. Give the machine more information so that it can do the job that you’re hoping for it to do for you.

Second is it does help to have frameworks. And the one that Kyle taught us is what he called RGCOA: Role, Goal, Context, Output, and Adjust.

So for instance, if you were trying to redesign your backyard, you could ask the AI to take on the role of a garden designer or a landscape designer. The next one, the G, is the goal. What is the goal? Because maybe if you’re just looking to explore, or what are your options, explore different costs for different materials. So depending on what the goal is, it’s a very different conversation. So be clear. What is the goal?

The next thing is the context. And that’s really the beef of your request. That’s got all the details. That’s where you would put pictures and dimensions and current state and preferences, and maybe examples of other things that you like or you don’t like.

Next is the output. What do you want the output to be? Do you want it to be a PowerPoint? Do you want it to be a PDF? Do you want it to be a bid?

And then the last thing, and this is so critical, ask it: do you need any more information for you to complete this task? And then what you’ll often find is it will ask you all kinds of stuff, clarifying questions. Maybe it’s setting your priorities. So always at the end ask, do you need any more information to get this thing done?

The triple shot. If you are going to invest in yourself, and you’re going to invest in these tools, and you want to learn them, it’s worth getting the paid versions. And really for me, and you can figure out what’s right for you, the big three are ChatGPT by OpenAI, Claude by Anthropic, and Gemini by Google.

What you can do is take your same prompt, copy paste it into each of the different gen AI tools, and you’ll get very different responses. If you’re going to get into this, pick something to buy, get the paid one. It’s like 20 bucks a month.

For me, HBO Max is probably the next one to go, to sacrifice that 13 bucks a month or 16 bucks a month, whatever, I don’t need. Also, each of them are good or better at different things. OpenAI is great voice. So when you’re walking around talking and you want to have a conversation, OpenAI on your phone works really, really well for that.

Although I will say, I was using Gemini earlier today, and Gemini using Google Voice is super, super powerful. The thing I love most about Gemini is anything that’s Google-related, search-related, finding thumbnails, finding references, finding, you know, kind of any little nugget that you get a piece of and you want to go back to the source. I find Gemini to be really, really well.

And as part of the Gemini suite is NotebookLM, and Notebook is cool because you can really easily add your own files. Same thing with Gemini. You can connect it directly to your Google Docs pretty easy as well, if you want to get it in there. Another reason to have the paid versions. You’re going to get much secure security on your data if you’re using these tools.

The thing though that I love about Claude, and the more I’m using Claude, the more it’s kind of my go-to, is it shows you its work. And even if you don’t exactly know how and exactly what it means in terms of coding, you can usually kind of follow along and see what the steps are, see how it’s kind of going through the logic of its decision making, which makes it really helpful when you want to go through—

Transcript (Conversation-Centric, Verbatim, Reformatted)

Jeff Frick:
…and interact and continue to change and update and modify what you’re doing.

So another thing that’s really important is managing tokens and managing context windows. So tokens is the language of gen AI. It’s about four characters, five characters, thereabout. It’s not a full word. And those are called tokens.

And tokens are what is counted when you put data into the context window, which is when you have your conversation with the gen AI. And you can blow it up. You can have too much and exceed the context window’s capability over the course of a conversation, over the course of a conversation.

So what you want to do then is often you will get to a point where you want to compress, or, you know, basically summarize, or get a new version or a shortened version, compression of your conversation so that you can continue it in potentially a new context window.

So managing context windows, managing tokens becomes a really important thing. And when you’re uploading or you’re reading attachments, there’s a tradeoff between whether you put that text copy-paste in the context window or do you attach a text file. And then when you attach a text file, is the gen AI reading the text file or is it keeping it there for future reference?

If you’re really starting to add more of your own data, which is when this stuff gets really fun, you can very quickly run out of those. And my example was just trying to upload a whole bunch of transcripts from 50 episodes of a podcast. It was just too much.

Specifically, there’s a function called temperature, which is the variability that the AI has in answering the question. You can adjust that lever, so you can make it more wild, you can tighten it up, but in fact it’s a feature, not a bug. Because when they designed these tools, they didn’t want it to always come back with the same answer every time. That’s why it’s gen AI. It generates a new answer that’s never been seen before versus a calculator.

So those are five things I learned with KMo at my AI class. Never stop learning. And it’s hard to keep up, but pick a few things to focus on. And for me that was Claude, Gemini, and to a lesser degree, OpenAI.

Thanks a lot. Take care.

Cold Close / Open Mic (Post-Goodbye):

Jeff Frick:
Hey, Jeff Frick here with a special shout-out to the podcast audience. Thanks for listening in. You can find transcripts and show notes at www.TurnTheLensPodcast.com. That’s one word, www.TurnTheLensPodcast.com.

Thanks for listening in. Share, subscribe, and smash that notification bell. Have a great day. Take care. Bye-bye.

‍