Episode 2512 New

Which AI Tool Should I Use?

ChatGPT? Claude? Gemini? Perplexity? When a solopreneur photographer asks which AI tool to use, it opens a bigger question: does it even matter? Bert and Julianna explore the "too many pens" problem and when specialization actually helps.

Episode Summary

A photography business owner asks: "Should I use ChatGPT or Gemini for this specific task?" The answer might surprise you. Julianna's advice: use what you have. But Bert digs deeper into when specialization actually matters - coding with Claude, writing with ChatGPT, images with Midjourney, academic research with Notebook LM.

From the 80/20 rule to the "too many pens" problem, this episode cuts through AI tool anxiety. Learn Bert's actual tech stack, when to pay for specialized tools, why Canva's AI counts as real AI use, and the critical difference between general-purpose models and product-specific AI wrappers.

Episode Timeline

00:00

The Question

A solopreneur photographer asks: "When should I be using each AI tool? Someone told me I need to use Gemini instead of ChatGPT for something." Julianna's initial advice: use what you have.

01:23

The 80/20 Rule

Bert introduces his philosophy: "If I can use 20% of the effort and get 80% of the result, I'm gonna do that." If you already have it and it's available to you, that's probably the one you should use.

02:00

Down the Rabbit Hole

Bert searches "what are your competitors?" across different AIs and finds ChatGPT, Claude, Gemini Advanced, DeepSeek (the Chinese entrance everyone freaked out about), Perplexity, and more. The list gets overwhelming fast.

03:00

Why Bert Chooses Different Tools

Julianna asks why Bert uses Claude for the website but ChatGPT for copy. The answer: some tools have differentiated themselves. Claude Code sits right in the IDE for coding. ChatGPT works great for writing. Different use cases.

04:00

I Gotta Ship Product

Bert catches himself: "I gotta be careful because this is a rabbit hole and I've got crap I need to get done today. I gotta ship product, man." The danger of over-researching AI tools.

04:24

Academic Rigor & Notebook LM

Bert discusses conversations with his wife and brother (both professors) about academic rigor. How do you ensure AI references are based on real documents? Enter Notebook LM - create your own RAG (retrieval augmented generation).

06:00

The "Too Many Pens" Problem

Bert collects fountain pens: "Yeah, I do have a lot of pens and sometimes I have to choose which one to use. And sometimes it's harder than it should be. A person with one pen knows which pen they're gonna use."

07:00

Writing & Token Preservation

Bert's strategy: Use ChatGPT for compelling writing because it's trained in his voice with fewer iterations. Also, he wants to preserve Claude tokens for coding work. It's about resource allocation.

08:00

Images: Midjourney vs The Rest

For images, Bert uses Midjourney. He's tried Google, GPT, and Claude for pictures, but "sometimes it just, I feel like it's an idiot when it comes to pictures." Julianna shares horror stories of AI changing her face.

10:00

Product-Specific AI

Julianna has specific AI tools for podcast editing. The conversation shifts to product-specific AI: Zoom has AI for meeting summaries, Canva has AI for image manipulation, tools built into products you already use.

12:00

Bert's Actual Tech Stack

The full list: Claude Code for coding, ChatGPT for writing/planning, Midjourney for generating images, Canva's AI for image work and backgrounds. "That's very tight. Subject to change because they're all evolving at a rapid pace."

13:00

Canva Counts as AI Use

Julianna highlights that many female solopreneurs already use Canva and wonder if they're "really using AI effectively." The answer: Yes! Asking Canva to remove a background or change a shirt color IS using AI effectively.

15:00

Judgment is the Human Skill

Bert: "Make sure that your judgment is good. What is going to actually solve the problem that I'm trying to solve? You need to understand what you're trying to solve, and you need to understand what good enough is."

18:00

You're Probably Already Paying

Should everyone pay for an AI tool? Bert's take: "You probably already are. Either you're paying for it or the company that you work for is paying for it, or it's bundled with your phone or it's bundled with some service."

19:00

Wrappers Under the Hood

Zoom's AI, Microsoft's AI - many are wrappers for ChatGPT or Claude APIs on the backend. Understanding this helps you evaluate whether you need another tool or if you're already using the same model through a different interface.

21:00

Deep Knowledge Benefits

Julianna predicts that in two years, having deep knowledge of how Claude or ChatGPT works will be hugely beneficial in large organizations. If Zoom, Notion, Canva all run on these models, understanding the base tool gives you an edge.

25:00

The New Hobby Analogy

Bert's insight: When you start a hobby (mountain biking, woodworking, fountain pens), you get your first tool. As you get deeper, you realize it doesn't meet ALL your needs, so your requirements shift. AI is the same - start with one, refine as you learn.

26:00

Five Things By Lunch

Julianna's action list: 1) Choose one AI tool and commit, 2) Explore what your company already provides, 3) Check product-specific AI in tools you use, 4) Actually start using it (don't stay on the fence), 5) Focus on your use cases - how will YOU use it?

Key Insights

1

It Matters a Little More Than You Think, But Not As Much As You Fear

Some AI tools have differentiated themselves for specific use cases (coding, writing, images, academic research), but for most people, using what you already have access to will get you 80% of the way there.

2

Product-Specific AI "Counts"

Using Canva's AI to remove backgrounds, Zoom's AI to summarize meetings, or your email client's AI to draft responses - that's all "real" AI use. You don't need to be in ChatGPT all day to be leveraging AI effectively.

3

Avoid the Research Rabbit Hole

The danger isn't picking the "wrong" tool - it's spending hours researching tools instead of shipping product. Pick one, start using it, refine your understanding, then choose specialized tools as you discover specific needs.

4

Judgment is the Human Superpower

The critical skill isn't knowing which AI to use - it's understanding what problem you're solving, what "good enough" looks like, and whether the AI's output meets your quality standards. Hone your judgment, not your tool collection.

Bert's Current AI Stack

Use Case Tool Why
Coding Claude Code + VS Code Sits right in the IDE, integrates seamlessly with development workflow
Writing & Planning ChatGPT Trained in Bert's voice, fewer iterations, preserves Claude tokens for coding
Voice Conversations Grok (in car) Integrated in car, hands-free voice interface
Generating Images Midjourney Higher quality output than ChatGPT/Claude/Google for image generation
Image Manipulation Canva AI Background removal, logo work, quick edits without learning Photoshop
Academic Research Notebook LM RAG for huge documents, ensures answers come only from provided sources

"Subject to change because they're all evolving at a rapid pace."

Memorable Quotes

"If I can use 20% of the effort and get 80% of the result, I'm gonna do that."

— Bert Carroll on the 80/20 rule

"I gotta be careful because this is a rabbit hole and I've got crap I need to get done today. I gotta ship product, man."

— Bert Carroll

"A person with one pen knows which pen they're gonna use."

— Bert Carroll on the "too many pens" problem

"There is this idea of like, oh, how can I use AI? What are these big, scary tools? But a lot of them are using Canva already."

— Julianna Fricchione

"Make sure that your judgment is good. You need to understand what you're trying to solve, and you need to understand what good enough is."

— Bert Carroll on human skills

"Staying on the fence and being afraid to start is hurting you more than actually engaging with it."

— Bert Carroll

Build Your AI Stack with Confidence

Join Automate to Accelerate - our comprehensive course on choosing the right AI tools for your work and integrating them into a sustainable workflow that actually ships product.

Be the first to know when enrollment opens