Agentic Commerce News for the Week of 11/2--11/8 (Week 45/52): Amazon vs. Perplexity and ACO on the Rise

Busy week in Agentic Commerce content out this week - 4 surveys, lots of perplexity news and more.

Only 20 days till Black Friday! 😱📈

At Retailgentic this week, we had a slower ‘big news’ week in Agentic Commerce:

• Tuesday - Instacart announced a wave of new AI offerings that we posited illustrates they are moving to more of a Merchant Services/unbundled model vs. the GSI model.

• Thursday - Since it was published on Sept 14th, I’ve had a ton of questions about the OpenAI paper, “How People Use ChatGPT” and it was awesome to get one of the authors, Ronnie Chatterji, Chief Economist at OpenAI, on the pod. He had a tight window of 30mins and I tried to squeeze as much out of it as I could. Watch it here, it’s a dense tight look at some great data and trends super-relevant to our Retail world and consumer behavior:

This Week’s Other Agentic Commerce News

Outside of those items we covered, it was an extremely active week with well over 20 interesting tidbits. We’ve narrowed that down to the ‘must reads’:

• Amazon had two big stories: an AWS deal with OpenAI and shots fired with Perplexity’s Comet agent!

• Perplexity had a very busy week

• ChatGPT showed some new ‘content cards’ and Shoptalk revealed their Spring focus area (you’ll never guess!)

• There were 2, count them 2, articles talking about Agentic Commerce Optimization

• An OG wrote an analysis of the impact of the search→ answer engine transition that 🤯

• Chinese retailers are doing some really innovating things using AI for singles day.

• Surveys, get…your….fresh….surveys - we have a 4-pack of new survey data!

• Finally, Shopify dropped a huge Agentic Commerce state in the mix of their Q3 earnings.

Let’s jump in…

Amazon News!

In last weekend’s News update, I went pretty deep into Andry Jassy’s conference call comments around Agentic Commerce and based on what Jassy said about “right exchange of value” read those tea leaves to mean we could see a bundled deal between OpenAI and Amazon like this:

If I’m Amazon I say, “Ok, we will put our content in your network but here’s a package deal:”

• 3P is at 4% rev share

• 1P is at 3% rev share

• You agree to a $100B AWS commit on Trainium3 for compute

• We get to invest $3B into OpenAI

• I’d ask for OpenAI to be avail on AWS as an API, but my understanding is Microsoft has this locked up, so probably not in the mix.

• You agree that if you build an ad network, it includes the Amazon ad network in its network so our ad buyers have access to your inventory and we get to sell against your ad signals/users

I also predicted we’d see Amazon on ChatGPT by July 2026.

Well imagine my surprise Monday when everyone started messaging me - “did you know this AWS deal was coming?” I was walking the dog and hadn’t checked news yet and turned on CNBC to this $38B deal between AWS and OpenAI.

Sadly, that’s all that was announced, but you could ALSO see: “oh we’ll do the chatgpt deal in a bundle and here’s how we’ll role it out:

• We’ll tease it in our Q3 conf call

• Then we’ll jointly announced the AWS piece

• Then on June 30, 2026 we’ll announce the ChatGPT piece so that Scot wins his bet with the readers of Retailgentic

Ok maybe that last bullet was a bit much, but let’s see how this unfolds! 🍿🍿🍿

Amazon vs. Perplexity: SHOTS FIRED

Then 🔥 Reuters announced Tuesday that Amazon had sued Perplexity over their Comet assistants ability to help users shop on Amazon. Sharp eyed retailgentic-ers may recall back in both our Comet and Atlas demos, we had this diagram that talks about how, unlike server-based crawling and even virtual machine based agentic models (standalone ChatGPT Operator→Agent) Amazon and other ‘BotBlockers’ would not be a able to stop this model of Agentic commerce. Game over.

FYI, this is because as a part of the browser, there’s no way for Amazon to discern if it’s human or AI and you really don’t want to get that wrong, so <game over>.

Well, 💩. Umm guys implied in my statement was the bold word here: “there is no TECHNOLOGICAL way to stop it”. That does not mean there isn’a potential legal issue here.

Perplexity 🥊 Back!

Perplexity’s CEO, Aravind Srinivas, is one of my favorite CEOs in the GenAI landcape because he is super 🌶️SPICY!🌶️ and he did not disappoint on this one:

Here’s the article.

Amazon v. Perplexity One We’ll be 👀 Closely:

We’ll be watching this very closely, here’s where we are today:

Amazon’s position:

Amazon claims Perplexity’s Agentic shopping tool is misusing Amazon’s platform, bypassing disclosure (doesn’t identify itself), undermines the user experience, and possibly violating the retailer’s rules, and therefore Amazon is enforcing its policies.

Perplexity’s position:

Perplexity defends Comet’s Agentic shopping feature as user-directed, transparent on its end, and claims Amazon’s threat is about protecting dominance rather than genuine concerns about transparency or user experience.

My non legal take:

Amazon is protecting their ad business here and I don’t buy that this degrades the user experience or eliminates personalization - I’m logged in ON Amazon, I’m getting all the personalized stuff, ahem, except spammy ad noise. 🤔

Perplexity News Roundup

While we’re on the topic of Perplexity, they had a busy week, enough to have a mini roundup for just their activities:

Perplexity Announced as the Default AI for Snapchat

Perplexity’s Comet Agentic Browser Coming to Android

Previously Comet was only available on Mac, now coming to Android! Video leaked here.

Perplexity Comet Also Coming to iOS?

Aravind tweeted ‘Very soon’ to a customer saying: ‘ios when?’→

Comet Receives Major Upgrade to Assistant

Perplexity to Add Kimi K2 Model?

There’s a new Chinese open-weight model, Kimi K2, by Moonshot (funded by Alibaba, so there’s an ecommerce connection!)

Aravind posted this shot of Kimi K2 crushing the standard benchmarks and said they are looking into adding it as an available model. What’s attractive is it is 60 cents per million token ‘in’, $2.5 1m token ‘out’. To put this in perspective, ChatGPT-5 is 2x ‘in’ at $1.20, and 4x ‘out’ at $10.

In other words, most applications will be 66-75% cheaper on a model like this. It came out today, it’s early, let’s see how it handles more testing and pushing. Sometimes these models overfit to the benchmarks and fail to perform the basics well.

ChatGPT Product Cards Showing as Info/Content Cards

I was doing some searches for news and imagine my surprise when <bloop> a product card popped up - wait, that’s not a product card. My 🧠 is so trained to see product cards, I was super confused. Turns out ChatGPT loves the ‘three card carousel’ concept so much (probably because it’s working with users in their engagement data) that they are now spreading it from it’s incubation area over in our product world out into the content world. I haven’t explored this too much in education, travel, or other areas, but wouldn’t be surprised by this next time next year there was a wide variety of cards popping up all over the place.

Shoptalk Announces Their Spring Show Theme and 🥁🥁🥁

This is great to see because I didn’t make it to either of the 2025 US shows, but the feedback from Spring was - it’s kind of interesting they didn’t have a lot of content around Agentic Commerce. Then in Fall I talked to folks and they were more like: “Can you BELIEVE they didn’t have more content about GenAI? Turns out, I guess they were saving it for the bigger show in the Spring. It’s going to be interesting to see, given their multi-tracks and multi-day content-filled agenda how they fill the content and segment the topic. I do think by Spring 2026, there will be more than enough surface area to cover.

Agentic Commerce Optimization (ACO>>AEO>>GEO>>SEO)

We introduced the idea of ACO in late October and this week there were two retail thought leaders writing about the concept.

Retail Media Breakfast Club

First we have Kiri Masters in her Retail Media Breakfast Club newsletter here.

Jason Cottrell Orium CEO + MACH President

Jason is a industry CEO and was just elected as the MACH alliance President. He posted this piece.

The Marketing <> Analytics Intersect - Avinash Kaushik

Avinash is the OG SEO/Analytics influencer/expert/thought leader in ecommerce as well as SEO. For 15yrs+ he was the Analytics Evangelist at Google, helping everyone understand Google Analytics, how to use it for checkout stage measurements, etc.

He’s been running this newsletter for years maybe decades called TMAI - it covers where marketing and analytics intersect and it has a huge readership in SEO and ecommerce.

For the last months he’s been covering the accelerating decrease in organic traffic and the increasing importance of AEO and it’s been great, but I do wish it was less SEO-like and more ACO, but he’ll get there 😀

Anyway, this week he casually drops in this statement right in the top of the newsletter:

“you are going to lose traffic (and revenue) somewhere in the range of 18-64%” - Avinash Kaushik

He then provides a well thought out model (available to his paid subscribes in XLS) that calculates the traffic and revenue lost for a $12m ecommerce company:

In this example (he admits he’s being conservative on the impact here), it comes out to a $4m revenue decrease, or 33%. I have a people a lot of SEO and SMB ecommerc people just had a ‘oh 💩’ moment this week when they realized how devastating the impact from these changes are going to be if they don’t take action very quickly to mitigate.

I’ve been thinking about this topic for >8 months and this article even freaked me out. In any case, I include it because the calculator is well thought out and if you haven’t done an impact analysis for your business, is an excellent framework.

Chinese Retailer AI Innovations

Forbes had an interesting article that goes uses the up-coming Chinese selling event of Singles day (11/11) to review some of the interesting Chinese retailer’s GenAI innovations. Several of them were things I haven’t seen and it’s worth a scan if this topic of ‘how to use AI on-site is interesting to you.

The one I wanted to highlight is here:

Based on recent prior purchases, Taobao provided Cheryl Huang with a wish list for 11.11 shopping that she said knew better what she wanted, than she did. It noticed she liked to make the dessert Tiramisu, so it recommended tools and baking pans. Because her Bichon Frisé is so tiny, she bought cat toys for it in the past. Taobao’s engine recommended cat toys it thought her dog would like. The product recommendations came with relevant promotional offers that might have been hard or time-consuming for her to search for, so she felt she was getting the best deal possible. The engine knew Cheryl had knee problems a few months back, so it recommended protective sleeves and relevant exercise equipment. Because the content that came back around Cheryl’s interests was so educational, relevant, and varied, it was fun and engaging to read.

In this example, because in China everything is a ‘super app’, within Taobao it’s the users core search engine, or answer engine now and their ecommerce marketplace. Therefore Taobao is able to contextualize all the users past informational AND product searches (not just a little sliver) and offer up a hyper-personalized and super-relevant shopping wish list.

While the intent of the article was an innovation retailers can learn from, I actually don’t think any retailer can provide this, there are really only four companies now that have successfully tied being your search engine and your shopping assistant: Google, ChatGPT, Perplexity and Microsoft Bing/Copilot.

(Survey)^4

It was a super busy week for Agentic Commerce oriented survey drops with four solid surveys that dropped. There’s so much information in here, we’ll hit 2-3 of the top insights and you can learn more in the links if interested.

Klaviyo

Klaviyo released their “Klavio AI shopping index” survey data here→

Smb Q4 25 Aiconsumertrendsreport Amer

7.6MB ∙ PDF file

Download

Download

Survey specs: n=3000, August 25, US/UK/AU, online shoppers (at least 3 purchases online in last 3 months)

Highlights:

• 81% have used AI for shopping or product research in last 3 months (highest I’ve seen now in the post ChatGPT-5 era)

• 56% plan to use AI for BFCM

Talkdesk

Talkdesk is a company that sells software for improving the helpdesk/support experience, so as you can imagine in their survey here, they ask a lot of questions on that topic which is not our scope. Survey specs: n=1000, US only, no timeframe given, report marked as late October. So imagine Sept/Oct survey time. They also have some retailer data and surveyed n=150.

Highlights:

• 75% of consumers surveyed plan to use AI to find deals, and 67% for gift ideas, signaling AI is now a holiday shopping essential.

• 84% of consumers surveyed say it’s important to know when AI is being used, but only 51% of surveyed retailers using AI disclose it.

• Among retailers surveyed who use AI, 80% expect higher holiday sales, and 79% anticipate stronger customer loyalty.

Worldpay

As their name implies Worldpay is a payment processing company. Their survey is here and the survey specs: n=8,000 total across US, UK, France, Brazil, China, Singapore and Australia, for US n=1,000

Highlights:

• $261B in Agentic Spend predicted - A strong headline stat that got me to the piece. In the teeny tiny print they explain that in the US survey respondents selected 9% of their purchases will be agentic in 5yrs so they took that 9% and multiplied it times $2.9T (a 2030 forecast) and get the $261B). Once we stack survey data on top of an unreferenced 2030 forecast, I’m out, but to be honest, I think its directionally way off, it’s going to be way way bigger. We’ll check in with you in 2030 to see how I did.

• 63% of respondents want help finding the lowest prices; 44% say they want help choosing top-rated brands

• 44% of Americans say they would let an AI assistant browse for them – rising to 59% among 18–34

• Only 6% say they would hand over complete control to an AI – most still want to review purchases first

Constructor

Constructor is an on-site search company, so, again, lots of questions on that topic in their survey here→

2025 State Of Ecommerce

4.24MB ∙ PDF file

Download

Download

Survey specs: n=1,500, US+EU.

Highlights:

• 64% have tried GenAI for shopping and 58% are comfortable with it

• 68% of consumers say site-search needs an upgrade

• 41% feel like strangers on a retailer site they use a lot (lack of personalization)

• 50% don’t like sponsored listing, 25% would pay to turn them off (YES!)

Anthropic’s updated Revenue Guidance

Anthropic is out with a new investor pitch that leaked and here’s a killer graphic that shows the bonkers revenue growth that both OpenAI and Anthropic are projecting to investors. By 2028, OpenAI looks to be north of $100B and Anthropic at $70B. I presume in the presentations there’s more meat on the bone of how they get there. Boy I’d love to see how much of that is Agentic Commerce! The rumors are that hot on the heels of OpenAI’s $500B valuation, Anthropic is going to try and raise in that same zone.

Shopify

Shopfiy announced their Q3 and had some interesting tidbits:

The headline is they said they are seeing an 11X spike in orders across their system from Agentic AI.

Next Week on Retailgentic

With only 20 days to Black Friday, my agentic Spidey Sense is telling me we’re going to get some very interesting and 🔥 news this week and next. Get some 😴 this weekend, it’s going to be a 🚀 week!