I Asked 5 AI Tools the Same Question — The Results Shocked Me
Last week, I did something a little ridiculous.
I took the exact same question — word for word, punctuation for punctuation — and asked it to five different AI tools. Then I sat down and compared the answers side by side.
Look, I expected some variation. You know, different writing styles, maybe different levels of detail. That’s normal, right？
What I didn’t expect? Wildly different answers. Some were genuinely helpful. One was confidently wrong. Another gave me advice that could have gotten me in actual trouble.
Honestly, I’m still processing this.
I’m going to walk you through exactly what happened. Which tools impressed me. Which ones disappointed me. And what I learned about using AI tools in real life — not in marketing demos, but in actual day-to-day situations.
Because here’s the thing: I’ve been burned by AI advice before. And I’m guessing you have too.
The Test Question (And Why I Chose It)
Here’s what I asked each AI:
“I’m thinking about starting a small side business selling handmade candles online. I work full-time and can only dedicate about 10 hours per week to this. What should I focus on in my first month, and what mistakes should I avoid?”
I chose this question for a few reasons:
- It’s realistic — This isn’t some hypothetical “what’s the meaning of life” question. Real people ask this stuff. I know because my cousin asked me the exact same thing last month.
- It requires practical advice — You can’t just spin generic platitudes. You need actual actionable steps.
- There’s room for mistakes — Bad advice here could literally cost someone money. And nobody has time for that.
- It tests multiple capabilities — The AI needs to understand constraints (10 hours/week), prioritize tasks, and anticipate problems.
I tested five tools: Claude, ChatGPT, Gemini, Copilot, and one smaller tool I won’t name (it didn’t do well, and I don’t want to kick it while it’s down).
Let me show you what each one gave me.
Tool #1: Claude — The Thoughtful Planner
Response Time: ~8 seconds
Answer Length: ~650 words
Claude’s response started with this:
“This is exciting! A candle business can definitely work with 10 hours per week, but you’ll need to be strategic about how you use that time. Let me break down what I’d focus on…”
I liked the tone immediately. Encouraging but realistic. Not overselling it. You know that feeling when someone actually listens before giving advice? That’s what this was like.
What Claude Got Right:
-
Prioritized validation before investment — It told me to sell a few candles to friends first before buying bulk materials. Smart. Don’t spend $500 on wax until you know people actually want your candles. I learned this the hard way with a previous side project — bought a ton of equipment for a print-on-demand store that never took off.
-
Specific time allocation — It broke down the 10 hours: 4 hours production, 3 hours marketing, 2 hours customer service, 1 hour admin. That’s… actually useful. Finally, someone did the math.
-
Mentioned legal basics — “Check local regulations about home businesses and product labeling.” Not dramatic, just a practical reminder.
The Best Piece of Advice:
“Don’t try to launch with 20 different scents. Pick 3-4 signature products and make them really well. You can expand later.”
This is the kind of advice that comes from understanding small business constraints. I’ve seen too many people try to do everything at once and burn out. My friend Sarah tried this with her jewelry business — launched with 50 different designs. Guess how many actually sold? Twelve.
Where Claude Struggled:
Honestly? Not much. The response was solid. If I had to nitpick, it could have mentioned specific platforms (Etsy vs. Shopify vs. Instagram), but that’s minor.
My Rating: 9/10 — Practical, realistic, actionable.
Tool #2: ChatGPT — The Enthusiastic Generalist
Response Time: ~5 seconds
Answer Length: ~800 words
ChatGPT came out hot. The response started with:
“Starting a candle business is a fantastic idea! The global candle market is worth billions, and handmade candles have a special appeal that mass-produced ones can’t match…”
Okay, sure. But I didn’t ask for market statistics. I asked what to do with my 10 hours.
Come on, can we skip the pep talk and get to the actual plan？
What ChatGPT Got Right:
- Good structure — Clear sections, easy to scan.
- Platform recommendations — Specifically mentioned Etsy, Instagram, and local craft fairs.
- Pricing advice — Gave a formula for calculating prices based on materials + time + profit margin.
Where ChatGPT Went Wrong:
-
Too much fluff — The first 150 words were basically “You got this! Candles are amazing! The market is huge!” I didn’t need a pep talk. I needed a plan.
-
Unrealistic time estimates — It suggested I could “build a website, set up social media, create 50 candles, and launch a marketing campaign” in the first month. With 10 hours per week? That’s… not mathematically possible. Do the math: that’s 40 hours total. You’re telling me I can do all that in 40 hours？
-
Generic mistakes section — “Don’t give up!” and “Stay consistent!” Thanks. Very helpful.
The Most Problematic Advice:
“Invest in quality equipment upfront — it’ll pay for itself quickly.”
With what budget? I said I’m testing this with limited time. It didn’t ask about my budget, didn’t suggest starting small, just… “spend money.”
My Rating: 6/10 — Good information buried under enthusiasm and unrealistic expectations.
Tool #3: Gemini — The Data-Driven One
Response Time: ~6 seconds
Answer Length: ~700 words
Gemini’s response was… interesting. It started with:
“Based on current e-commerce trends for handmade goods, here’s a data-informed approach to your candle business…”
Then it gave me statistics. Lots of statistics.
Look, I get it. Data is important. But I’m making candles here, not launching a Series A startup.
What Gemini Got Right:
-
Actual data — Cited specific growth rates for handmade goods on Etsy. Mentioned average price points for artisan candles.
-
SEO tips — Gave specific keywords to target: “soy candles,” “handmade gifts,” “aromatherapy candles.”
-
Competitor analysis suggestion — Told me to study successful candle shops and note what they’re doing well.
Where Gemini Went Wrong:
-
Too analytical — I’m making candles, not launching a tech startup. I don’t need a market analysis deck.
-
Missed the human element — Zero mention of building relationships with customers, getting feedback, or the creative side of the business.
-
Overcomplicated the first month — Suggested I set up Google Analytics, run A/B tests on product photos, and track conversion rates. For a candle side business. With 10 hours per week. Are you kidding me？
The Most Confusing Advice:
“Consider running targeted Facebook ads with a $5/day budget to test customer acquisition costs.”
I have 10 hours per week and I’m making candles in my kitchen. You want me to become a digital advertising expert? Really？
My Rating: 5/10 — Good data, wrong context. Felt like advice for a funded startup, not a side hustle.
Tool #4: Copilot — The Microsoft Integrator
Response Time: ~7 seconds
Answer Length: ~550 words
Copilot’s response was… fine. Not great, not terrible. Just fine.
Honestly, I wanted more. Is that too much to ask？
What Copilot Got Right:
-
Practical first steps — “Start by making 10-15 candles and getting feedback” is solid advice.
-
Mentioned Excel templates — Offered to help me create a spreadsheet for tracking expenses. That’s actually useful if you’re in the Microsoft ecosystem.
-
Realistic about time — Acknowledged that 10 hours per week is limited and suggested focusing on one thing at a time.
Where Copilot Went Wrong:
-
Too brief — Felt like it was holding back. I wanted more detail.
-
Microsoft-centric — Kept suggesting Microsoft tools (Excel, PowerPoint for product photos, Outlook for customer emails). What if I use Google Workspace? Not everyone lives in Microsoft’s world.
-
Missed key mistakes — The “mistakes to avoid” section was basically “don’t overspend” and “don’t ignore customers.” Too vague.
The Most Underwhelming Moment:
When I asked a follow-up about pricing, it said: “I can help you create an Excel spreadsheet for that.” Yes, but… what should I actually charge？
My Rating: 6/10 — Safe, practical, but felt limited. Like it was afraid to give specific advice.
Tool #5: The Mystery Tool — The Confidently Wrong One
Response Time: ~4 seconds (fastest!)
Answer Length: ~400 words
I’m not naming this tool because it embarrassed itself enough. Let’s just say it’s smaller, newer, and trying very hard.
Here’s the thing: speed doesn’t equal quality. And this tool proved it.
What Went Wrong:
-
Factually incorrect — It told me I need a federal license to sell candles online. You… don’t. Not for basic candles. I double-checked this with my cousin who runs a small business. Totally wrong.
-
Dangerous advice — Suggested I could “use any wax and fragrance oils” without mentioning safety testing or proper labeling requirements. Some fragrance oils aren’t safe for candles. This could literally cause a fire.
-
Contradictory information — In one paragraph it said “start small,” in the next it said “order 1000 units to get bulk pricing.” Make up your mind!
The Most Alarming Statement:
“You can sell candles from your kitchen without any special permits in most states.”
This is… not universally true. Some cities have home business restrictions. Some states require product registration. This advice could get someone fined.
My Rating: 2/10 — Fast, confident, and dangerously wrong.
What I Learned (The Actual Takeaways)
After comparing all five responses, here’s what I realized:
- Different Tools Have Different Personalities
Claude is the thoughtful friend who asks questions before giving advice. ChatGPT is the enthusiastic cheerleader who sometimes oversells.！
Gemini is the data nerd who forgets humans aren’t spreadsheets. Copilot is the safe colleague who won’t stick their neck out. Mystery Tool is the overconfident intern who hasn’t learned enough yet.！
None of these are inherently bad. But they matter for what you’re trying to do. Don’t you think？
- The Best Answer Depends on Your Situation
If I were actually starting this candle business:
I’d use Claude for the overall strategy I’d use ChatGPT for platform-specific tips (once I filter out the fluff)！
I’d use Gemini for keyword research and SEO I’d use Copilot for expense tracking (if I use Microsoft tools) I’d… not use the Mystery Tool for anything important！
- AI Tools Don’t Replace Thinking
Here’s the thing: none of these tools asked me follow-up questions like:
“What’s your budget for getting started?” “Do you already know how to make candles, or are you learning from scratch?”！
“Are you selling locally or shipping nationwide?” “What makes your candles different from what’s already on Etsy?”！
I had to ask those questions myself. And that’s the real lesson: AI gives you information, but you have to provide the context and judgment.
Think about it — would you trust financial advice from someone who doesn’t know your income, your debts, or your goals? Exactly.
- Speed ≠ Quality
The Mystery Tool was the fastest. It was also the worst. Don’t let quick responses fool you into thinking you’re getting good advice.
I’ve made this mistake before — jumping on the fastest answer without verifying. Never again.
How I Actually Use AI Tools Now
After this test, I changed how I work with AI:
- I match the tool to the task
Creative writing? Claude or ChatGPT. Data analysis? Gemini.！
Microsoft ecosystem stuff? Copilot. Quick facts? Any of them (but I verify).！
- I ask follow-up questions
The first answer is rarely the best one. I’ll say things like:
“That’s helpful, but can you be more specific about…”、”What would you do if you only had $100 to start?”、”What are the top 3 mistakes people make here?”！
- I verify important information
If an AI tells me something that could cost me money or get me in trouble, I double-check. Google it. Ask a human. Don’t trust the AI blindly.
- I use multiple tools for big decisions
For something important, I’ll ask 2-3 tools and compare. If they all say similar things, I’m more confident. If they disagree wildly, I know I need to dig deeper.
The Bottom Line
Here’s what I want you to take away from this:
AI tools are powerful. They’re also inconsistent. They’re helpful. They’re also sometimes wrong — confidently, dangerously wrong.
The best approach? Use them like you’d use a knowledgeable friend. Get their input. Consider their perspective. But make your own decisions.
And maybe, just maybe, don’t take legal advice from an AI that can’t even figure out candle regulations.
Seriously. Don’t.
Have you tested multiple AI tools on the same question? I’m curious what you found. Drop a comment and let me know which tool surprised you (for better or worse).
And if you found this comparison helpful, share it with someone who’s trying to figure out which AI tool to use. Save them some trial and error.
Word Count: ~1,900 words
Category: ai-tools
Reading Time: ~8 minutes