Amazon Support Couldn't Fix It. ChatGPT Did in About 30 Seconds. | Edition 321
Edition 321 — A $191 battery backup looked completely dead. Amazon said contact the manufacturer. ChatGPT fixed it in one test.

I had one of those little experiences this week that perfectly explains why I think AI is going to become part of everyday life.
Not because it wrote a poem. Not because it generated an amazing video. Not because it built some complicated piece of software.
It helped me fix something in my house. And it did it faster than traditional customer support.
My Almost-New UPS Was Dead
On June 14, I bought an APC Back-UPS Pro BR1000MS battery backup from Amazon. It cost about $191.
It’s the type of device you plug your computer and electronics into so that if the electricity goes out, the battery keeps everything running for a while. It also provides surge protection.
I plugged it in when I received it. Everything worked great.
About a week later, I was leaving on vacation for an extended period, so I unplugged it. When I came home, I plugged it back in.
Nothing.
It appeared completely dead.
I pushed the power button. Nothing.
I checked the wall outlet. Good.
I checked the plug. Good.
I checked the battery. I checked the reset. I pushed buttons. Still nothing.
And I’m thinking: how does a nearly brand-new $191 UPS die while sitting there unplugged?
I Contacted Amazon
Since I had bought it from Amazon, I contacted Amazon support. They couldn’t really help troubleshoot it. The purchase was already outside the normal return window, so they essentially told me: you’ll need to contact APC.
APC is the manufacturer. From Amazon’s standpoint, that probably made sense. But it didn’t solve my problem. I now had another company to contact, another support process to go through, and probably more time spent explaining what had happened.
Before I did that, I decided to try ChatGPT.
I Asked ChatGPT to Help
I explained exactly what had happened: I bought the UPS June 14. It worked perfectly. I unplugged it before vacation. I returned and plugged it back in. It wouldn’t turn on. I had already checked the obvious things.
ChatGPT suggested we determine whether the problem was with the battery or the UPS itself.
There was one important instruction I gave ChatGPT: “One at a time.”
I didn’t want a giant troubleshooting checklist. I wanted one instruction. I’d perform it. I’d tell ChatGPT what happened. Then it could decide what to do next.
That turned out to be the perfect way to troubleshoot it. Four tests later, the UPS was working again — the full sequence:
| Step | What ChatGPT Had Me Do | What Happened | What It Proved |
|---|---|---|---|
| 1. Disconnect the battery | Unplug the internal battery, plug the UPS directly into the wall, press power | It turned on and beeped | The UPS's main electronics weren't dead |
| 2. Reconnect the battery | Unplug the UPS, reconnect the battery firmly, plug back in, turn on | It turned on normally | Reconnecting the battery fixed something |
| 3. Pull the plug | Leave it running, unplug the UPS from the wall | It stayed on and beeped, as designed | The battery itself worked |
| 4. Power a real device | Plug a lamp into a battery-backed outlet, then unplug the UPS | The lamp stayed lit | It could actually carry a load |
What Fixed It?
We can’t say with absolute certainty what happened. But there are a couple of likely possibilities.
Disconnecting and reconnecting the battery may have restored a connection that wasn’t making perfect contact. Or disconnecting the battery may have reset the UPS electronics after it had been sitting unplugged.
The important part isn’t really which one it was. The important part is how ChatGPT approached the problem. It didn’t guess. It tested.
Can the UPS work without the battery? Yes — that told us the electronics were alive. Does it work with the battery reconnected? Yes. Will the battery actually power the UPS when electricity disappears? Yes. Can it support a real device? Yes. Problem solved.
And This Took About 30 Seconds to Figure Out
That’s the part I keep thinking about. Amazon support couldn’t troubleshoot it for me. Their process got to: past return period → contact manufacturer.
ChatGPT approached it completely differently: let’s figure out what’s actually wrong. One test. One answer. Next test. Next answer. And very quickly, the problem was solved.
I didn’t need to return anything. I didn’t need to buy another UPS. I didn’t need a replacement battery. I didn’t need to spend an hour on the phone with APC. My $191 piece of equipment was fine.
This Is What People Miss About AI
Most AI news focuses on the spectacular stuff. AI can generate video. AI can write software. AI can create music. AI can analyze huge amounts of information. AI can build agents. All of that is interesting.
But I’m becoming convinced that some of the most valuable AI applications are going to be much simpler. Something isn’t working — ask AI. You don’t understand something — ask AI. You don’t know what something is — take a picture and ask AI. You’re about to call customer service — ask AI first.
That’s where AI starts becoming less like a website you occasionally visit and more like an assistant that’s available whenever you need it.
One of My Favorite AI Tricks: Take a Picture
This has become one of the easiest ways to use AI. You don’t even have to know how to describe the problem. Just take a picture.
You can photograph an appliance, an error message, your television screen, your computer settings, a piece of equipment, a control panel, something in your car, a menu, a sign, a plant, a bug, an unfamiliar object.
Then ask: “What am I looking at?” Or: “What does this mean?” Or: “Which button should I push?” Or: “What’s wrong here?”
That’s very different from Google. With Google, you often have to know what you’re searching for. With AI, you can simply show it the problem.
Before You Call Someone, Ask AI
I’m starting to use this as a rule. Before I call technical support, the cable company, an appliance company, an airline, a hotel, a service technician, or a manufacturer — I ask ChatGPT first.
Sometimes ChatGPT solves the entire problem. Other times it helps me understand the problem before I call. That’s valuable too.
Instead of saying “it doesn’t work,” I might be able to say “the device works on AC power, but the battery isn’t carrying the load.” That’s a much better support call.
When You Can’t Figure Something Out
This might be one of the most useful everyday applications of AI. We’ve all had this experience: you’re trying to change a setting, something isn’t working, you can’t find an option, you don’t understand the instructions, you spend 20 minutes clicking around.
Now I increasingly stop and say: “I can’t figure this out. Walk me through it one step at a time.”
That last part matters. One step at a time. You don’t always want 17 instructions. You want: do this. Then tell the AI what happened. Now the AI has new information. It can decide what you should do next. That’s exactly what happened with my UPS.
I Use the Same Idea When I Travel
AI has also become incredibly useful while traveling. Before a trip, you can ask: “Help me plan a week here.” But it becomes even more useful once you’re actually there.
You can say: “I’m standing here right now. What should I see nearby?” Take a picture of a building: “Tell me about this.” Photograph a menu: “Explain every item and tell me what I might like.” Driving somewhere? “What interesting stops are along this route?” Have two free hours? “I’m here and have two hours. What would you do?”
That’s almost like carrying around a personal tour guide.
Explain This to Me
This may be the AI feature I use most. I read something. I don’t completely understand it. So I ask: “Explain this to me in plain English.” Then I keep asking questions — about technology, investing, economics, medicine, history, politics, contracts, instruction manuals, news stories, AI products.
Here’s what’s different from reading an article: I can interrupt. If I don’t understand something, I can ask another question. If the answer is too complicated: “Make that simpler.” If I want more depth: “Explain that part more.” If I want an example: “Give me an example.”
The explanation changes based on what I understand. That’s incredibly powerful.
And Sometimes I Just Ask for Ideas
You don’t even need a problem. Sometimes I start with: “Give me some ideas.” Ideas for a vacation, a business, a marketing campaign, a newsletter, dinner, something to buy, improving a product, solving a problem, things to do this weekend.
Then you start talking. One idea leads to another. Sometimes AI’s biggest contribution isn’t giving you the answer. It’s helping you discover questions and possibilities you hadn’t considered.
My New Rule: Ask AI First
This little UPS incident reinforced a habit I’ve been developing. Before I call someone, ask AI. Before I search for 20 minutes, ask AI. Before I replace something, ask AI. Before I return something, ask AI. Before I give up, ask AI.
And if there’s something in front of me that I don’t understand: take a picture. If I’m confused: “Explain this to me.” If something isn’t working: “Troubleshoot this with me one step at a time.” If I’m going somewhere: “Help me plan this.” If I don’t know what to do: “Give me some ideas.”
And sometimes the simplest prompt is still one of the best: “What would you do?”
Try This for One Week
Here’s my challenge for you. For the next week, notice whenever you find yourself thinking: “I don’t know.” “I don’t understand this.” “I can’t figure this out.” “I wonder what this is.” “I need to call someone.” “Something isn’t working.”
Before doing anything else, open ChatGPT. Ask. If you can show it what you’re looking at, take a picture. And if you’re troubleshooting something, try this exact prompt:
“Help me troubleshoot this one step at a time. Give me only one thing to try. Wait for me to tell you what happened before giving me the next step.”
That’s essentially what fixed my UPS. And the more I use AI this way, the more I think this is where the real revolution is happening. Not in some futuristic laboratory. Right in the middle of ordinary life. One little problem at a time.