We just wrapped up the 7 Figure Japan Mastermind in Tokyo.
2 days. Workshops, hot seats, and a lot of AI.
My biggest lesson from the room: AI is only as good as the context brain you give it.
This week we're covering my top 10 takeaways from Tokyo, Amazon's new AI seller assistant, 2 Black Friday deadlines coming up fast, and why video is driving TikTok Shop sales.
And I have a quick question for you: what do you want to build next?
Let's get into it.
My Top 10 Takeaways from the 7 Figure Japan Mastermind
Here's something that went wrong.
It was 3 p.m. the day before the event. I was getting everything ready for our dinner.
That's when I realized the name tags weren't ready.
Previously, my former assistant either made them herself or sent them to the printer.
So I used Codex, running GPT-6 Astra.
I took a picture of the name tags left over from our previous event. Then I had Codex build new ones based on that picture and the logo designs in my context brain. (Codex designed those too.)
Then it wasn't sure about the size.
I didn't have a ruler. It was a holiday at my co-working space. Nobody had a ruler.
So I took a picture of the name tag next to my keyboard and had Codex figure it out.
It got it right on the first try.
The first design was a little bland. So I had it make it better using our brand assets.
This is what it came up with:
Built in about 15 minutes with Codex and our context brain.
Not too bad, considering I literally did it in about 15 minutes.
And the print size fit the badge perfectly.
That's why the context brain is #1 on my list.
We covered a lot in Tokyo: building a business brain, AI for product images and video, PPC, and connecting agents to your business data.
On our alumni call yesterday morning, I went back through what mattered most.
Here are my top 10 takeaways you can take back to your business:
Make your business brain the foundation. Keep your product facts, brand rules, decisions and current work in one place that your AI tools can use. Without context, AI gives you generic answers.
This is how I run my own business now. My products, people, decisions and even my writing samples live in one folder that my AI tools can read.
Use one chief of staff, plus focused specialists. One agent keeps the overview. Specialist agents handle single jobs like PPC, customer service or research. Use a stronger model for hard decisions and a lighter model for routine work.
Connect your data, then check what's still manual. An MCP connection or your regular reports can feed a dashboard. But mark the gaps, and have a person check before you act.
Put a brand book in your brain. Your logo, colors, fonts, tone and examples, all in one place. Then your images, emails and videos all use the same references. As Chris Rawlings put it: "You just upload the brand book as a PDF with the prompt."
Turn plain phone photos into product-image references. Chris Rawlings takes about 12 raw phone photos of every product: different angles, close-ups, and something for scale. Those become the inputs for AI images.
Build AI video in stages. Can Chan's approach: study a strong video in your category, write your own concept, approve the still images, then make the video. Review scale, shape and labels at every step.
Use a fast product-validation loop. Concept, AI mockups, a split test with your target customers, revise, test again. It's a lot cheaper to change a mockup than a container of inventory.
Check your profit margin before you develop further. Start with your selling price. Subtract product, packaging, freight, fees, returns and ads. If there's nothing left, faster content won't fix it.
Make team AI adoption concrete. One person, one useful workflow, a demo at the next team meeting. Then share what worked.
Tune your agents, and keep your tool mix flexible. A specialist agent needs a clear brief, examples and feedback. Tools change fast. Don't marry one.
Chris Rawlings presenting at the 7 Figure Japan Mastermind in Tokyo.
Can Chan presenting at the 7 Figure Japan Mastermind in Tokyo.
TAKEAWAY: Don't try to do all 10 at once. Pick one workflow, give it one owner, and measure the result. Start with your business brain, because everything else gets better when your AI has the right context.
At Accelerate on September 23, Amazon announced new workflows for Seller Assistant, its AI assistant inside Seller Central. Early seller reports this week say it's now showing up in US accounts.
Here's how it works.
You describe a task in plain words. The assistant drafts the steps for you.
Nothing runs until you approve it.
Once approved, a workflow can run on a schedule, and you can see a history of what it did.
Early reports list starter workflows for things like an orders and returns digest, pricing checks, account health monitoring, and review requests. You can also upload your own business documents so it has more context.
Early reports also mention an offer: 12 months of Amazon Quick Plus free for the main account holder plus 2 users, if you sign up by December 31, 2026. Check the terms in your own account before you count on it.
So, is it any good? Honestly, I want to hear from you.
If you've tested it, hit reply and give it a score from 1 to 10. 1 = it sucks. 10 = it's amazing, I can fire my VA. Tell me what you used it for. I'll share the results in a future issue.
What I like so far is the setup: it drafts, you approve. That's the right way to start with any AI agent.
However, these are the risks.
It only knows what's in Seller Central and what you upload. It won't know your factory lead times, your cash position, or what's happening in your other sales channels.
TAKEAWAY: Test it with one low-risk workflow first, like a daily orders and returns summary. Check its output against your own numbers for 2 weeks. One workflow, one metric, one human approval.
Black Friday Deadlines: October 20 and October 28
2 dates to put on your calendar this week.
First, CruxFinder reports that Amazon's Black Friday / Cyber Monday deal submission window closes October 20. Amazon also listed an October 20 deals deadline in its Seller Forums post on peak fees.
Second, inbound cutoffs.
According to PPC Land and Be Bold Digital, minimal-split FBA shipments need to arrive by October 21, and Amazon-optimized split shipments by October 28. Your cutoff depends on the placement option you pick in Send to Amazon.
TAKEAWAY: Before you submit a deal, run the numbers after peak fees. A deal that sells more units but makes no money isn't a win. Check the inbound cutoff in your own account before you book freight.
Marketplace Pulse just released its TikTok Shop Trends Report 2026, built from more than 150,000 US sellers and 4.5 million products.
The big finding: the 8% of sellers with more than 1,000 videos account for 60% of estimated sales. The 39% of sellers with 50 or fewer videos account for just 5%.
What does this mean? On TikTok Shop, the number of videos you can put out matters more than the number of SKUs you have.
TikTok isn't the only place. Amazon listings, Meta ads and your own website all want more video too.
That's why everybody wants to do more AI video right now.
But nobody wants AI slop: weird hands, products that change shape between shots, and videos that look fake.
Customers notice.
And there are so many tools out there. How do you decide what to use?
Here's how I'd approach it, based on what we covered at the Japan Mastermind:
Start with the job, not the tool. A product demo, a UGC-style clip, and a 6-second ad each need something different.
Use your real product as the reference. Shoot photos from several angles and feed those into the AI.
Break it into stages. Concept, images, short clips, then editing. Review each stage before you move on.
Test 2 tools on the same shot. Judge them on how accurate your product looks, how much each usable clip costs you, and how long it takes.
Keep a human in the loop. Check shape, scale, labels and colors before anything goes live.
TAKEAWAY: Video volume matters, but bad AI video can hurt your brand. Start with one product, one video type, and one tool. Trust but verify.
Quick question: What do you want to build next?
Speaking of AI video: it's one of the options in a quick survey I want you to take.
There's a lot changing right now. AI video, second brains, PPC, inventory, and Q4 is right around the corner.
I want to make sure the next trainings I put together are the ones you actually need. So I put together a quick 5-minute survey.
What do you want to build next? For example:
AI video for your products or brand
An AI second brain for your business
AI PPC optimization
Inventory forecasting
Preparing for Black Friday / Cyber Monday
Or something else
Fill it out and you'll get a chance to win a free 60-minute AI assessment with me ($500 value). We'll look at your business together and map out the AI workflow that makes the most sense for you to build first. I'll announce the winner on Friday, October 30.
Take the 5-minute survey here: https://newsletter.7figuresellersummit.com/forms/57b97578-2dc8-4903-a845-444f781f3353
I read every response.
In case you missed it:
That's your 80/20 for this week.
Talk soon,
Gary
P.S. The survey takes 5 minutes. Opt in at the end for a chance at the free 60-minute AI assessment. I'll announce the winner on Friday, October 30.