I Asked an AI Dropshipping Agent to Research a Niche. It Told Me No.

I'm up to about 25 AI agents running real parts of my business. Maybe eight of them actually earn their keep. The rest are chatbots wearing a costume.

So when Doba sent me Doba Pilot, their new AI dropshipping agent, and told me it runs your whole store from research to live sales, I sat down and tried to catch it lying to me.

It didn't lie. But there's a wall in the middle of it, and most people won't see it coming.

Test one: can it actually research?

Research is where most AI tools quietly fall apart. They tell you what you want to hear. So I gave Pilot's Market Scout something deliberately broad: pet supplies. That's not a niche. That's an aisle. A weak tool takes that and hands you back a list of dog bowls.

It thought for about two and a half minutes, and I could watch what it was doing the whole time. Then it came back with a verdict I did not expect: cautious entry.

Here's why it said that. In US pet supplements, Amazon itself sells 70% of the units. Nutramax holds 45% of the revenue. Products that have been on the shelf longer than three years account for 81.6% of sales. The top 10 brands account for 100% of the sampled units, which means anyone outside that group is making basically nothing.

It narrowed to dog joint supplements and showed me 47,800 searches with a 7% purchase rate. That part is good. A 7% purchase rate means those people aren't just browsing, they're buying. But the supply and demand ratio came back at 10 competing products per searcher. I'd feel a lot better about 5 to 1.

None of that is a pitch. That's a tool telling me the door is mostly closed, and then showing me the receipts.

I gave the research an A plus, and I trust it more because it didn't tell me I'd be rich.

Test two: can it actually source?

New chat. Find me dog joint supplements I can sell for $24.99 to $34.99, show me the supplier and show me my margin.

First pass it only found one product, and nothing in my range. So it broadened the price filter on its own and came back with 16.

The thing that matters here isn't the products. Anyone can find products. These came with a supplier attached, and with the numbers I actually care about: how long that supplier has been in business, their fulfillment rate, and their refund rate. I don't want a supplier that opened three months ago. I want a 100% fulfillment rate and a refund rate at or near zero, because a 5% refund rate means the product they're shipping isn't good.

That information used to be brutal to dig up. Back before any of this existed, you'd find a product you liked and then spend days trying to work out who could actually ship it without scamming you. Pilot just handed it to me, with a 35% to 55% margin calculated against the price range I asked for.

Then it hits the wall

Research, done. Products, done. Suppliers, done. Next step is a store.

So I said list these products. And it said: you'll need a Shopify store, do you have one or do you need a new one?

This is the wall. Doba Pilot does not build the store. It does most of the work around it, but you go to Shopify yourself, start the trial, pay them separately, and then Pilot takes over from there. It even hands you the button to start the trial, so it's not hard. But if the words "autopilot dropshipping" put a picture in your head where you type into a chatbot and a finished store falls out the other end, that picture is wrong, and I'd rather you hear it from me than find out after you've paid.

What it costs, and who it's for

Every message costs one credit. A Market Scout report costs five. The 7-day trial is $0.99 for new accounts and gives you 30 credits, so you can run six full market reports inside the trial before you commit to anything.

This is a good fit if you've decided you want to sell online, you don't know what to sell yet, and the thing stopping you is that you don't have days to spend on research. It skips that entire step.

It's a bad fit if you're looking for a button that prints money. I've been looking for that button most of my life and it doesn't exist. As I said in the video, hard work brings profit, mere talk leads only to poverty. Anything worth doing is going to take real effort, and the people who never accept that just hop from one supposed shortcut to the next and never actually build anything.

The test I'd run

Point it at the niche you're already thinking about.

If it warns you off, you just saved yourself months of trouble. If it finds you a gap, you walk away with a supplier list and fulfillment rates attached. Either one of those is worth more than a tool that agrees with you.

Try Doba Pilot: https://www.doba.com?mtic=MT-GNRUCN
7-day trial for $0.99 (new accounts). Code FIKbVPDgqYvv gets you 10% off a paid plan. It doesn't apply to the $0.99 trial.

The follow-up test, where I left the agent running on a schedule for a week and came back to see what it had done, is written up in I Left an AI Agent Alone for a Week.