T. Brian Jones is co-founder and CTO of Inventory Hero. He leads the engineering behind its Amazon data pipeline, demand forecasting, and the AI platform that lets sellers talk to their live inventory, sales, and supplier data in plain language.
Legitimately, it means using software to handle the repetitive parts of running your own FBA business: repricing, requesting reviews, generating reports, monitoring stock and flagging reorders, and assisting with advertising and listings. You stay the operator; the tools remove manual labor. It does not mean a hands-off, passive-income business someone else runs for you. That second meaning is where most of the danger is, because 'Amazon automation' has become a term heavily associated with done-for-you scams.
Are Amazon automation businesses a scam?
Many of the 'done-for-you' passive-income versions are. The FTC has brought multiple cases against Amazon-automation operations that charged tens of thousands of dollars promising to build and run profitable stores, where the results did not materialize, stores got suspended, and in at least one case the FTC said the promised AI technology did not even exist. Legitimate seller tooling that you operate yourself is a completely different thing. The red flags are upfront fees in the thousands, guaranteed passive income, and someone else running your account.
What parts of an FBA business can I safely automate?
The mature, low-risk tasks are repricing (staying competitive within price floors and ceilings you set), review requests through Amazon's official Request a Review mechanism, and reporting plus restock alerts. With human oversight you can also automate a lot of PPC (rule-based bidding is reliable; fully algorithmic bidding still drifts around launches and seasonality) and listing drafts (AI generates good first drafts that need a compliance and accuracy review). The pattern is that repetitive, rules-friendly work automates well; judgment does not.
What should never be fully automated in an FBA business?
The decisions that carry real consequences: product and sourcing choices, supplier negotiation, committing cash to a purchase order, compliance and account-health judgment (appeals and plans of action), and exception handling. These need human judgment, and AI reliability degrades on long, open-ended tasks anyway. The safe operating model is human-in-the-loop: let automation draft the work and flag the issues, and keep yourself on the moves that spend money or affect your account.
Read article
Amazon FBA automation, done right, means using tools to handle the repetitive parts of running your own business, not paying someone to run a store for you. The short version: repricing, review requests, reporting, and restock alerts automate safely today, PPC and listings automate well with oversight, and sourcing, negotiation, and cash commitment still need you. First, though, a warning: "Amazon automation" is also a term heavily associated with prosecuted scams, so separating the two is the most important thing this article does.
The word "automation" hides two nearly opposite things, and conflating them is how people lose money.
The legitimate meaning is using software to remove repetitive labor from a business you run: a repricer, an alert system, a reporting tool, an AI assistant that drafts your reorders. You are still the operator; the tools just do the tedious parts faster than you can.
The dangerous meaning is the "done-for-you" or "passive income" pitch: pay a company a large upfront fee and they will build and run a profitable Amazon store on your behalf. The FTC has brought multiple enforcement actions against operations like this, including cases where stores were suspended en masse and where the agency said the promised AI technology did not actually exist. The tells are consistent: thousands of dollars upfront, guaranteed returns, and someone else controlling your account. If you see those, walk away.
The genuinely mature automations share a shape: repetitive, rules-friendly, and reversible.
Repricing. The most solved task. A repricer keeps you competitive within the floors and ceilings you set, so you are not manually chasing the Buy Box.
Review requests. Use Amazon's official Request a Review mechanism (tools that automate it are fine); custom external emails or incentives violate Amazon's terms, so stay inside the official channel.
Reporting and restock alerts. Automated performance summaries and low-stock warnings so nothing drifts unnoticed.
These run with minimal babysitting because the rules are clear and the downside of a mistake is small and reversible. Repricing is the clearest worked case: set a floor at your landed cost times your minimum margin, say an $8.50 cost plus $4.75 in FBA fees is $13.25, and a floor of $13.25 times 1.15 is about $15.24, so the repricer competes for the Buy Box but never sells below $15.24 no matter what a competitor does.
Two categories work well but need a human watching:
PPC. Rule-based bidding (raise or lower bids under conditions you define) is reliable. Fully algorithmic "set and forget" bidding still drifts around product launches and seasonal swings, so keep an eye on it, especially at those moments.
Listing generation. Amazon's own AI listing tools and third-party generators produce good first drafts, but they still need a human accuracy and compliance review before they go live. A confident draft with a wrong spec is worse than a blank field.
Amazon has been building seller-facing AI too. Its assistant (launched as Project Amelia and since expanded into a broader Seller Assistant) can answer questions, scan account health and compliance in the background, and suggest actions like discounting slow movers. Importantly, it takes those actions only with your approval, and richer capabilities are still rolling out. It is a useful signal of where things are heading, but confirm what is actually live in your Seller Central rather than assuming the marketing describes today.
Some parts of the business are not automatable in any trustworthy way, because they are judgment, not process:
Product and sourcing decisions. What to sell and who to buy from.
Supplier negotiation. Terms, MOQs, and relationships.
Committing cash. Approving a purchase order is a financial decision you own.
Compliance and account health. Appeals and plans of action need a person.
Exception handling. The weird, one-off situations where AI is least reliable.
There is a technical reason beyond caution: AI and agent reliability degrades on long, open-ended, multi-step tasks because errors compound (more on that in AI agent for ecommerce). So the correct model is human-in-the-loop: automation drafts and flags; you approve the consequential moves.
Start with repricing and alerts. Immediate time savings, minimal risk.
Add reporting and restock automation so your data watches itself and your reorders come pre-drafted.
Layer in PPC and listing assistance with a review step.
Add an AI assistant over clean data to answer questions and draft work.
Keep approval on everything consequential.
That gives you most of the leverage of automation without handing your business (or your money) to someone promising passive income. For the assistant side, see the AI employee for your FBA business; for the inventory piece specifically, AI inventory management.