Joe Nilsen

Machine Learning Engineer 🤖 Revolutionizing E-Commerce with AI

The AI conversation in e-commerce has been hijacked.By people who've never shipped a product. By "experts" who discovered LLMs last year and now talk like they invented them. By creators recycling the same surface-level garbage with a fresh screenshot.They're loud. They're everywhere. And they're getting real operators burned.I've spent 15 years in the Amazon trenches. Built and exited two brands. Architected the systems behind TraceFuse, Seller Strategies, and Selletek.Now I deploy AI inside real operations while the influencers are still workshopping their next thread about it.I'm not here to add noise. I'm here to end it.What I share is the operational reality of AI that nobody else will show you. What to automate and what to never touch. Workflows that hold up under real volume. The mistakes that sink teams. What's working now versus what's already dead.No fluff. No theory. No "exciting updates coming soon."If it isn't deployed, it doesn't exist.

Founder

Our vision is to make the world a more connected place through the art of cooking and passion for food. The platform was launched in 2015 and was built upon challenging the status quo. Over the years, Culinary Connect evolved into a movement celebrating cooking, food, and love. These core principles are the cornerstone of what the modern-day brand is built upon.

"Ex-Amazon" is not a credential. It's a timestamp.

Amazon in 2018 is not Amazon in 2026.

Different internal tools.
Different policy interpretation.
Different enforcement behaviors.
Different incentives.

Some of these people haven't worked there in 8, 10, or even 12 years. The Amazon they knew doesn't exist anymore. It's not even close.

But they learned something valuable while they were there. Not the systems. The language.

They learned to talk like insiders. How to throw around internal terminology and reference escalation paths that sound specific enough to be believable. How to say "that's a Seller Performance workflow" with enough confidence that operators assume they're still plugged in.

And they never stopped using it. That's the play.

Every consultation. Every webinar. Every LinkedIn post. Same phrases. Same insider tone. Same confidence. Spinning it just enough to sound current when nothing about their knowledge actually is.

Most operators can't tell the difference because they've never been inside the building themselves.

So the language works. Until it doesn't. 

Until the advice fails because it was built on a version of Amazon that hasn't existed for years.

Amazon rebuilt its enforcement automation. AI is embedded in workflows that were fully manual when these people had badges. Rufus changed how customers find products.

The catalog system was gutted and rebuilt. The review process is unrecognizable compared to even three years ago.

They missed all of it. Every single shift. Because they weren't there.

So when someone uses "former Amazon" as the center of their authority, here's what that actually means:

They're selling the brand, not the signal.
They're anchoring advice to a company that no longer exists.
They're renting credibility instead of producing current results.

If your edge is a past employer, you don't have an edge.

Sellers don't need mythology.

They need answers that work this week.

Founder & Operator @ MajestIQ | AI Systems, E-commerce

elderly man sitting in a rocking chair on a sunlit porch. He is wearing a faded orange 'Amazon Class of 2010' t-shirt and leaning forward with a cane, telling an intense story. Three young children sit on the floor looking up at him with expressions of confusion and boredom. The scene blends photorealism with caricature to create a 'back in my day' moment about corporate tech culture.

Amazon is now hiding reviews from buyers.

I just went to check one-star reviews on a product listing. Standard operating procedure. Nothing unusual.

Instead of reviews, I got a blue info box.

"We're showing a limited selection of reviews. To see more reviews, you can send a request."

Read that again.

Amazon, the company that built its entire marketplace on transparent buyer feedback, is now telling shoppers they need permission to read reviews.

I tested this across multiple listings. Same result. Try to dig past the curated selection and you hit a wall.

This isn't a bug. This is deliberate. 

For sellers, the implications hit immediately.

Your negative reviews are no longer visible to most buyers. Sounds great until you realize your competitors' negative reviews are hidden too. The entire competitive landscape around social proof just shifted overnight.

For buyers, it's worse. Someone spending $50 on a product can't do basic due diligence without submitting a formal request. The friction alone will stop 99% of people from ever seeing the full picture.

Amazon has spent two decades telling the world that customer reviews are sacred. That transparency drives trust. That the review ecosystem is what separates them from every other marketplace.

Now they're gating it behind a request form.

No announcement. No explanation. No public acknowledgment.

The company that built its empire on customer trust just told customers to submit a request to see it.

Screenshot showing how Amazon is now gating reviews and no longer allowing people to read them. 

I’m happy to share that I’m starting a new position as GTM and Partnerships at Parcelis!

Well, this sucks.

Or does it?

Amazon just dropped a new Agent Policy inside the BSA update. Effective March 4.

If an AI agent or automated system touches Amazon Services, it must:
- identify itself as automated
- comply with the policy at all times
- cease access immediately if Amazon requests it

That is not a “policy update.” That is Amazon drawing a line in the sand for automated access and AI tooling.

The conference circuit “AI gurus” charging $3,000 a ticket to teach “AI-Powered Ecom Domination” are going to have some explaining to do.

The “AI automation” agencies are about to find out the difference between a demo and a durable system.

Good.

This is what happens when people build AI on vibes instead of constraints.

They assumed unlimited access.
They ignored guardrails.
They treated policy like a suggestion.

The weekend Lovable vibe-coding ninjas who duct-taped an “Amazon AI tool” together and started calling themselves experts are about to get their skirts pulled up.

No authentication framework.
No API alignment.
No governance.
No kill switch.
No audit trail.

Just a pretty UI, brittle workflows, and wishful thinking.

A lot of these businesses were never building infrastructure. They were renting risk.

Now Amazon has the language, the process, and the enforcement path to shut automated systems down on demand.

This is a massive advantage for operators who built the right way from the start.

- Compliant architecture.
- Proper API access.
- Systems designed to survive policy tightening.

The new restrictions on using Amazon materials for AI development and the added protection against reverse engineering will wipe out a chunk of the current tool stack.

If your "AI" was built by having ChatGPT generate prompts for Claude Code or a similar tool, it has an expiration date.

Amazon didn’t kill AI in e-commerce.

They killed the fake version of it.

A screenshot of Amazon's new business solutions agreement outlining how Amazon is going to treat AI agents and how all sellers must comply with these new policies.

Two months ago, I exposed the review removal scam industry on this platform.

Since then, more than 15 sellers have reached out to me with horror stories.

One reached out last week, which is why I’m writing this post.

He paid a scam artist “negative review removal” service over $24,000 in the last six months.

When I walked him through the audit process from my original post, his vendor went silent.

No case IDs.
No pre-removal documentation.
Nothing.

Because there was nothing.

Amazon removed those reviews.
His vendor just watched for the 404 errors (dog pages) and sent invoices.

This is not a gray area.
This is fraud.
And it’s worse than I laid out the first time.

Here's the part that makes it nearly impossible to catch:

Amazon’s review enforcement is getting more aggressive, more automated, more proactive.

That doesn’t make these vendors harder to run. It makes them easier.

Because Amazon is doing the clean-up at an industrial scale, every single day, and these people sit downstream collecting “success fees” from removals they did not cause.

The scam is simple:

→ They scrape your branded listings
→ They detect reviews that are already dead (404, dog page, broken link)
→ They send you a “win” report
→ They invoice you $250 to $500 per “removal”
→ You pay them for Amazon’s enforcement

They are not removing anything. They are watching Amazon remove it, then billing you like they caused it.

And the scam gets nearly invisible due to account-level enforcement.

When Amazon bans a buyer account for manipulation, every single review that account ever posted gets wiped.

Not just the bad ones. All of them.

Your removal vendor did not cause that. Amazon did.

But your vendor will invoice you for it anyway.

The problem is not performance-based pricing.

The problem is charging you without proving causation.

If you are currently paying a vendor for review removals, do this today.

Pull every invoice from the last 6 months.

For each “removal” they billed you for, demand:

1. Case ID created BEFORE the review disappeared
2. Timestamped confirmation the case was opened before removal
3. The policy violation cited by Amazon
4. Amazon’s response confirming removal due to that case

Not a screenshot after the fact.
Not a dead link.
Not a report showing the review is gone.

You already know it’s gone.
You need proof they made it go away.

For the last time, I'm shouting it from the hilltops...

Stop paying for the rain and thanking someone for their dance.

Infographic exposing Amazon review removal scam. Shows a 5-step fraud process: 1) Scrape listings, 2) Monitor deleted reviews showing 404 dead links, 3) Send "win" report, 4) Invoice the seller, 5) Collect "success fees" of $250-$500 per "removal." Highlights that Amazon's automated enforcement removes reviews at scale while vendors take credit. Bottom section shows red checkmarks next to "No Case ID, No Proof, No Documentation" with the message "They did not remove anything. Amazon did. They just took the credit."

I was not expecting to cry at my desk before 7 AM today.

Eric Dane died three days ago. ALS. He was 53.

Before he passed, he filmed an episode of Netflix's Famous Last Words. Part of it was directly to his daughters. Four lessons from a disease that was slowly taking his body.

I watched it this morning, and I couldn't move.

Not because it was sad. It was.

Because I felt like he was talking to me.

1. Live in the present.

He said he spent years lost in his head. Worry. Self-pity. Shame. It took a terminal diagnosis to force him into the now. And once he got there, he didn't want to leave.

"The past contains regrets. The future remains unknown. The present is all you have. Treasure it."

2. Fall in love with something.

Not someone. Something. Find the thing that makes you want to get up in the morning and hold onto it.

3. Choose your friends wisely.

Find your people. Let them find you.

"They don't do anything special. They just show up."

4. Fight with dignity.

The disease was taking his body. It was never going to take his spirit.

"Fight until your last breath. Fight, girls, and hold your heads high."

I move fast. I stay locked in on work. That's who I am. I'm not changing that.

But I watched a dying man tell his daughters to treasure the present, and I had to sit with the fact that I don't do that well enough.

I lose sight of what matters more often than I want to admit.

I'm carrying this with me.

More presence.
More intention.
Kinder to the people around me.
Less time in my own head running the next scenario.
More time actually being where I am.

Not because a LinkedIn post told me to.

Because a father's last words made it impossible to ignore.

Five-minute video worth checking out: https://lnkd.in/ekJ3tPPe

Eric Dane speaks directly to his daughters in a final recorded message, sharing four hard-earned lessons from living with ALS: live in the present, fall in love with something, choose your friends wisely, and fight with dignity.

I got on a call at 8 pm last night fully expecting to be off in 20 minutes.

An hour later, I was still on the phone. Fired up. Notebook full of ideas. More energy than I had at 9 am.

The founders of Prompt Cowboy from Fourday wanted to connect with some of their users. I've been using the platform daily for a while. Had my whole team sign up and adopt it. So, I took the call.

Two co-founders. Both remarkable. The kind of people who are quietly building something serious while everyone else is out here posting about AI they've never deployed.

The conversation took many turns:

- Real AI pain points in actual operations. 
- Context management challenges that every operator knows, but nobody writes about. 
- The future of how these tools need to work versus how they work today. 
- Ideas bouncing back and forth.

I walked away reminded of something fundamental: The single best thing you can do as a builder is spend time with other builders.

Not content creators talking about building.
Not consultants advising on building.
People who are actually in the trenches, building.

That energy is contagious. It reignites the obsession. It pulls the best thinking out of you. 

Every operator I know who stays sharp over the long haul has this in common. They stay connected to people building at a high level.

This call was a perfect example. I showed up tired. I left inspired.

About the platform itself: Prompt Cowboy is significantly more than a prompt tool. The team had the foresight to build around context management and features that operators actually need as AI usage matures.

It's powerful. And from everything the founders shared, it's about to get a lot more powerful.

Do yourself a favor by checking the platform out.

I have no affiliation. You'll be glad you did.

Screenshot of the prompt cowboy platform shows how you can build different prompts and have the prompt type selection. It's in dark mode, a very powerful platform. 

taplio