What we're listening to
Notes on the seller podcasts we follow, written in our own words and shared with each show's permission. Every episode links to the full show. Newest first.
-
Smart Problem Solving for Amazon Sellers: Fix 100 Issues at the Root
Instead of working through a hundred problems one at a time, this episode suggests stepping back to find the handful of conditions behind most of them. Group the issues, make an educated guess about what drives each group, check those guesses with whoever knows the business best, then try one small change and see whether several problems ease at once. In the real case behind the episode, a friend's long list of AI issues, the approach turned roughly ten days of work into two to three hours.
Our take: Account health tends to work the same way. A run of warnings on one account usually comes from one place, and the fix that lasts is the one that removes it.
-
Running Claude and Codex From Your Phone For Amazon Sellers
Back for the monthly show, the guest admits her AI tools now move faster than she does, which makes her the bottleneck on her own projects. The answer discussed here is steering the work from a phone: plan properly at the desk, then handle questions and corrections while out walking or cooking. The pair cover the catches of each remote option, where AI usage really adds up, how to keep a large library of reusable skills tidy, and why protecting attention beats chasing every release.
Our take: Stepping away is easier when the work is set up to run without you. On an account, that means knowing which alerts need a person the same day and which can safely wait.
-
Fable 5.1 vs Astra: For Amazon Sellers
Two major AI models launched within 72 hours, and this conversation argues that asking which one is better misses the point. A model is only the engine; what decides results is the setup around it, from the job and the prompt to the tools, how failures are handled and the standard an answer has to meet. The guest also explains why two models that look equally cheap on paper can end up with very different bills, and why polished demo images say little about everyday business work.
Our take: The same test applies to anything written for your account: judge it against what Amazon actually asks for, not against how polished it reads.
-
Cutting AI Costs Without Cutting Output For Amazon Sellers
A brand owner with a 25-person team on Claude compares notes with the host on keeping that spend under control. Their working split: a stronger model for planning, a cheaper one for routine building, and big projects broken into stages so each session stays manageable. They also explain why hopping between providers often costs a trained team more than it saves, and finish with a better question than what AI costs: what is it actually producing for the business?
Our take: When an account is in trouble, the cost that matters most often never shows on an invoice: it is the days the account isn't selling.
-
Signal to Noise: AI Inbox Systems for Amazon Sellers
Host and guest tackle the same problem from different sides: AI speeds everything up, and the inbox fills with noise. The guest shows two systems she built, one that quietly files unsolicited pitches before she sees them, and one that reads years of client emails to flag which relationships are warming or cooling. The host describes his own short, labeled morning routine for clearing mail. Their shared conclusion is that firm boundaries now matter more than adding yet another tool.
Our take: Amazon's most important messages often land in the same inbox as the noise. Make sure performance notifications and account health warnings are never filtered away with the pitches.
-
RAG for Amazon Sellers and Claude Model Switching
When most of what a business knows sits in the founder's head and a stack of spreadsheets, every question lands on one desk. This episode explains retrieval augmented generation (RAG) in plain terms: break your own documents into pieces, index them, and let an AI answer from them with sources attached, set up to admit when it has no answer. The second half covers sending routine AI work to cheaper models, and why AI video editing still needs a person who knows the craft.
Our take: Well-kept records pay off long before any AI reads them. When Amazon asks for invoices, supplier documents or a timeline, the seller who can find them in minutes is already ahead.
-
Cognitive Overload: AI Maxing, Product Development and the Hidden Tax
The first of a new monthly roundtable, recorded as the show nears its tenth year. One guest walks through product development run largely with AI, from scoring ideas on keywords to generated concepts and renders, with a launch pipeline booked into 2027. Another makes the case that deciding what comes first is the most underrated skill right now. The second half turns to mental load: checking AI output all day, constant switching between tasks, and why attention, not time, is now the scarce resource.
Our take: Decision fatigue shows up in account health too. Under pressure, sellers tend to send the fast appeal rather than the right one, and that is often when a case starts to loop.