Friday means Q&A. I pulled the most common questions from this week’s emails and comments. No fluff. Just straight answers to the stuff people are actually wondering about but afraid to ask.
Let’s go.
Q: “WHICH AI MODEL SHOULD I USE: CHATGPT, CLAUDE, OR GEMINI?”
A: Wrong question. Use all three for different things. ChatGPT (specifically GPT-4o) is fastest for quick tasks and has the best plugin ecosystem. Claude Sonnet 4.5 is best for complex reasoning and long-form content. Gemini 3.0 is best for processing large amounts of data and multimodal work (images, video, audio together).
Here’s what I actually do: use Galaxy.ai to access all of them through one interface. Route tasks to whichever model is best for that specific job. Stop treating this like a religion. It’s just tools.
Q: “HOW DO I KNOW IF AI IS ACTUALLY SAVING ME TIME OR IF I’M JUST PLAYING WITH TECH?”
A: Track it. Seriously. Before you implement any AI tool, write down how long a task currently takes. Then time it with AI. If you’re not saving at least 50% of the time within two weeks, you’re doing it wrong.
Most people waste time on AI because they’re automating things that don’t matter. Focus on your three most time-consuming repetitive tasks. Automate those first. Ignore everything else until those are running smoothly.
TACTICAL TAKEAWAY #1: THE TIME TRACKING METHOD THAT PROVES ROI
Do this for one week:
1. Pick your three most repetitive tasks (email, research, content creation, whatever)
2. Track exact time spent on each task for three days before AI
3. Implement AI solutions
4. Track exact time spent on each task for three days with AI
5. Calculate savings in hours per week
6. Multiply by your hourly rate
If the math doesn’t work, you’re using AI wrong. Fix it or drop it.
Q: “I TRIED BUILDING AUTOMATION WITH MAKE.COM AND IT BROKE AFTER TWO DAYS. WHAT AM I DOING WRONG?”
A: You didn’t build in error handling. Most automation breaks because people assume everything will work perfectly forever. It won’t.
Add fallbacks. If an API call fails, route to a backup. If data is missing, flag it for human review instead of crashing. Build notifications so you know when something breaks before your clients do.
Also, test with bad data. Don’t just test the happy path. Feed your automation garbage inputs and see what happens. Fix the failures before they matter.
Q: “IS IT WORTH PAYING FOR AI TOOLS OR SHOULD I JUST USE THE FREE VERSIONS?”
A: Depends on volume. If you’re sending 10 AI requests a day, free is fine. If you’re running real business operations, you’ll hit free tier limits in about 3 hours and then you’re dead in the water.
Here’s my rule: free versions are for testing and learning. Paid versions are for making money. If an AI tool is actually saving you time or generating revenue, pay for it. The subscription cost is noise compared to the value.
Exception: if a tool has a free tier that covers your actual usage, milk it. But don’t cripple your operations trying to save $20/month.
Q: “HOW DO I GET MY TEAM TO ACTUALLY USE AI INSTEAD OF IGNORING IT?”
A: Build it into their existing workflow, don’t make it extra work. If you tell people to “start using ChatGPT,” they’ll nod and then never touch it.
Instead, build automation that handles parts of their job automatically. They wake up to emails drafted, research compiled, reports generated. They edit instead of creating from scratch.
Make it invisible. Make it helpful. Make it impossible to ignore because it’s saving them hours every week. Then they’ll ask how to use it more.
TACTICAL TAKEAWAY #2: THE TEAM ADOPTION PLAYBOOK
WEEK 1: Pick one repetitive team task everyone hates. Build AI automation for it. Deploy without asking permission.
WEEK 2: Track time saved. Share results in team meeting with actual numbers.
WEEK 3: Offer to build similar automation for other tasks. Let people volunteer their pain points.
WEEK 4: Train interested team members on the tools. Share templates and workflows.
Adoption happens when people see results, not when they sit through training. Lead with value.
Q: “WHAT’S THE BIGGEST MISTAKE YOU SEE PEOPLE MAKING WITH AI RIGHT NOW?”
A: Treating AI outputs as final instead of first drafts. The people failing with AI are the ones who copy/paste without thinking. The people winning are the ones who use AI to get 80% of the way there, then add the human touch that makes it actually good.
AI is a force multiplier, not a replacement. It makes good people great and mediocre people slightly less mediocre. If you don’t know what good looks like in your domain, AI won’t fix that.
Q: “SHOULD I LEARN PROMPT ENGINEERING OR JUST USE TOOLS THAT HAVE PROMPTS BUILT IN?”
A: Learn the basics, use tools for everything else. Spend two hours understanding how prompts work: be specific, give examples, define output format, set constraints. That covers 90% of use cases.
For complex stuff, use tools like Claude’s prompt generator or existing prompt libraries. Don’t reinvent wheels when someone already built the car.
THE ONE TOOL EVERYONE NEEDS
Multiple people asked about my “essential AI stack.” Here’s the honest answer: you need one automation tool and access to multiple AI models. That’s it.
AUTOMATION: Make.com. Connects everything, visual interface, doesn’t require coding. Start here.
AI ACCESS: Galaxy.ai. One subscription, all major models, easy model switching. Cheaper than paying for each model separately.
Everything else is optional based on your specific needs. Email tool, CRM, project management, whatever. But those two tools let you build AI workflows that actually work.
Want the complete setup guide with workflows, prompts, and step-by-step videos? It’s in the AI BUSINESS ACCELERATOR.
👉 COMMENT “FRIDAY” and I’ll send the details.
NEXT WEEK
Monday: The new AI video tools that don’t suck
Tuesday: How to use AI for market research without getting garbage data
Wednesday: Latest model updates and what they mean
Thursday: Building AI workflows that don’t break
Friday: More Q&A
Same time, same place. Have a good weekend.
Jordan Hale
The AI Newsroom
Seattle, WA
Real AI. Real Results. No Hype.