Hey,
It’s Pete Matheson with a new issue of Experiments in Progress.
This week is about AI and the trap of trying too many tools.
🤖 The AI That Actually Helps (And What Just Added Noise)
Most AI feels impressive when you first try it.
There’s so much hype, and new tools appear constantly. It’s very easy to fall into the trap of trying all of them.
But the real question is:
Is it actually helping… or just adding more noise?
I ran into this recently with fitness tracking.
I’ve worn:
a smartwatch
a smart ring
a Whoop
sometimes two watches at once
Each one gave slightly different data. Instead of helping, it created doubt.
Instead of clarity, I was checking four apps and thinking about it more.
AI can create the same problem: more tools, more switching, more friction.
📊 What each AI is actually best at
After using most of them regularly, this is how I’d summarise where each one fits:
Tool | Best for | Avoid using it for |
|---|---|---|
ChatGPT | Thinking, explaining ideas, comparing options | Deep analysis of large personal datasets |
Claude | Analysing documents, emails, large context | Quick factual searches |
Gemini | All-round use, research, multimedia, Google integration | Precise reasoning or nuanced decisions |
Perplexity | Fast research, finding sources, replacing Google | Deep workflows or ongoing projects |
Bixby | Simple on-device phone actions | Complex questions or research |
Microsoft Copilot | Integrating with Office, documents, Windows workflows | General-purpose thinking or creative work |
✅ What I’d recommend depending on your level
Most people don’t need multiple tools.
Start simple.
Level | Recommendation | Why |
|---|---|---|
Beginner | Use ChatGPT only | Best balance of capability and simplicity |
Intermediate | Add Claude | Extremely useful when analysing documents and information |
Advanced | Add Gemini or Perplexity | Useful for research and broader workflows |
Everything else is optional.
You don’t need six assistants. One or two used properly is enough.
⚙️ Action steps: how to figure out what actually helps
If you want to understand the differences quickly, try this:
1. Give each AI the same task
For example:
Ask ChatGPT, Claude, and Gemini the same question
Compare how they answer
Notice which one feels most useful
You’ll quickly see they think differently.
2. Use Claude on your own data
Try:
uploading a document
analysing notes
reviewing emails
This is where it becomes much more powerful.
3. Use Perplexity instead of Google for a week
It’s often faster and more direct. Especially for research.
4. Pick one tool and go deeper
AI is genuinely useful but only when it reduces decisions.
My experience is that the biggest improvement didn’t come from adding more tools. It came from using fewer and using them properly.
Which AI tool do you use most right now?
📱 Enjoying this newsletter? Share it with a friend who’s as obsessed with tech as you are:
I’ve now spent enough time using the 52" Dell UltraSharp to answer the questions that matter, especially the ones you only discover after living with it.
Inside the video:
Exactly how I connect multiple computers and run the full setup
Compatibility questions (Mac, PC, Linux, refresh rates, GPUs)
Whether ghosting, noise, or performance issues show up in practice
How it compares to OLED and other high-end monitor options
The accessories and setup choices behind the scenes
40" vs 52": which actually makes more sense for productivity