Agentic Shopping News for the Week of 8/10-8/16 (week 33/52): Theme this week: Agentic Commerce Goes Mainstream
Some new Agentic datapoints, Great reads from: A16Z+Curious Commerce+Ian Jindal, Target makes moves, Agentic shopping goes mainstream with consumer pubs and legacy old-school ecommerce vendors.
Welcome to our weekly curated recap of news around GenAI and Agentic shopping that we think is most relavant to retailers, brands and other companies in the digitial retail ecosystem.
This week the bulk of the GenAI press was around the ChatGPT-5 announcement and some of the fallout and immediate changes OpenAI had to do. The TL;DR - turns out, unlike traditional software, consumers from a relationship/bond with these models and when you just take one out behind the barn and, well you know, they freak out, so they brought back good ole ChatGPT-4o for folks.
Zuckerberg’s META Earnings Subtle Comment Surfaces
I’ve been building in the software world for 30yrs now and never seen anything like what we’re seeing as we build ReFiBuy. You may have picked up a bit on what we’re seeing and ‘feeling’ from my Reinforcement Learning with AI Feedback article here. The headline is: for the first time, we are going to see software that self-improves and ‘gets smarter’ over time - without any code updates. We’re seeing evidence of that at ReFiBuy and it’s something I would have never thought possible and, to be honest, it’s a bit weird to get your head around.
I didn’t have a chance to listen to the Meta conference call so missed this, but this week, a full 2 weeks later, top AI people I follow started asking: “Wait, did we hear Zuck say the have ASI (Artificial Super Intelligence) now?!”
They are referring to this segment:
To me, interpreting this as ‘Meta has ASI’ is a stretch. What I believe happened is Meta created an Agentic system that uses RLAIF for their ad systems where it can run infinite combinations of creatives against very large audiences with clear goals. Unlike traditional algorithms or data science that is static, the Agentic system can get smarter and adapt at inference time. That doesn’t mean that the system was writing new code as many implied, although my idea of ‘what is capable now’ has been so expanded in the last 3 months that I no longer would be shocked if that happened either. Most likely Meta now has created an ad-optimizing Agentic+RLAIF system that’s improving itself without any ‘human in the middle’ oversight or direction.
What the heck does this have to do with Digital Retail?
Fair question! The reason I mention this concept is every brand, merchant, seller, retailer and agency has a complex technology stack.
We are now entering an era where the ecommerce enablement agentic software you use is going to ping you and say: “Good Morning Sarah, I noticed our bounce rate was up, so I reorganized the home page and found a better optimized version.” or “we sold out of product x and I saw product y was heating up, so I went ahead and took six actions based on those datapoints last night at 3am. Oh yeah and GenAI engine X made a change, so I added that to all the SKUs in the catalog and updated all our channel partners.
This ‘Agentic-ification’ not going to happen overnight, but in < 5 years the digital commerce job will look more like collaborating with a team of agents that learn and share their findings with you. Shall we call it Vibe Commercing? Too Early?
Last thought on this topic to ponder over your weekend: We’ve even seen a couple of our more cheeky agents (one generates data and one checks the data/performs QA - in what’s called an eval config in Agent-land) get in some pretty heated disagreements that can go on for quite a while until they settle it. It’s pretty surreal to go read the transcripts of those conversations.
New GenAI/Agentic Shopping Datapoints this Week
Long-time readers know I love a good fresh GenAI/Agentic datapoint and this week had some doozies.
First, at a high level this is a good illustration of how fast revenue-wise the two leading frontier LLM providers OpenAI and Anthropic:
That curve from 2024 ($6.5B) to 2025 (18B) at around 3X is unlike anything we’ve ever experienced before at this scale. As if that wasn’t crazy enough, ChatGPT projects $26.4B for 2026 and Anthropic is $10B - put those together for $37.4B which is a 2X.
2027 is a stretch, but have:
OpenAI: $86-100B - $93B at the mid-point
Anthropic: $12-$35B - 18B at the mid-point
For combined mid-points of $111B - back to 3X 🤯🚀🚀
SEMRush Projected Visitors/Source (Trad Organic vs. LLM)
SEMRush came out with a banger chart that projects the current trends in traditional organic search (blue) against the growth in GenAI (red) with a ‘Total’ trend line' (green).
You can see SEMRush is projecting:
An overall drop in total traffic to websites
The drop levels-off around 2027
GenAI traffic crosses organic in 2028
A lot can happen to change the trajectory of either line, but it is helpful to have a line in the sand we can start measuring going forward.
Prediction: I’m taking the ‘way under’ on the 2028 lines cross. We should setup a Polymarket on this 🧐.
Burke: Shopper Behavior Analysis
Burke is a research firm that focuses on understanding consumer behavior, the shopper behavior and how to optimize the buyer’s journey. Their Shopper Scientist, James Sorensen, has a great piece out this week you can find here.
Standout data from the report:
82% of consumers that used GenAI for shopping, found it made the journey faster
They have some great charts that look at generational adoption, dig into the issue of trust and much more.
I though this was particularly interesting, it’s the first ‘categorical breakout’ of GenAI shopping that I’ve seen.
CNET: Agents now make up a third of all traffic to Brands
CNET had an article out sourcing an SEO consultancy, Brightedge, that has data indicating that up to a third of all brand traffic is now coming from GenAI If you want the full report it’s here.
Great Reads This Week: A16Z, Curious Commerce and Ian Jindal
I call this section, Great Reads - three pieces that really stood out this week as thoughtful, new takes and really add to the conversation.
A16Z Proactively Asks: “Is Google Screwed?”
Ya’ll know I love a good framework. Melina at
, introduced us to her ‘Value x Vibes’ framework
In the same vein, In this awesome post from A16Z, Justine Moore and Alex Rampell think through the amount of consideration the shopper makes for a purchase from low (impulse buy) to high (life purchases). They then matrix that across the pre-internet ‘analog era’ to ecommerce 1.0 to Agentic and give us another framework.
What’s interesting to me about this line of thinking is that I tend to get in comment discussions with industry people that say things like: “I’d use an agent to reorder stuff, but never groceries and certainly not a high considered purchase.”
I understand this reaction, but I think they can’t imagine anything but the ‘order for me’ scenario working. But I think the way you interact and leverage the ‘Agentic superpower’ will be different across the consideration contexts. For example, with the daily essentials, it will ‘just order it’ and save you time. But on a higher-consideration purchase, it will be like your GPS, helping you navigate the Research part of the funnel and maybe you stop there and go solo for the rest, or you decide to tell it ‘ok, I decided on X and want you to find it for $Y” and alert me to what you find.
A16Z digs deep into the concept and framework and it’s a great thought experiment of where Agentic is going for the consumer experiences and how Agents will adapt to the shopping journey to fill different supporting roles.
If you want to really work your Agentic thinking, see if you can map Melina’s four quadrant Value x Vibes framework to the A16Z framework… or, vice versa.
Speaking of great thought experiments…
Melina @: Loyalty in the Age of AI
Melina Flabiano is back this week with yet another great piece. This time she digs into thinking through what loyalty will mean to brands and retailers in the Agentic Shopping Era and I particularly thought you’d enjoy the ‘Future-proofing loyalty’ programs section that is actionable and the pontificates on what future AI-native loyalty programs could look like. The last part made me want to start a company in that space, but dang it my dance card is full already.
Across the Pond, Ian Jindal Reminds Us: Customers First, Even in Agentic
Ian Jindal is a UK-based Digital Expert who has been in digital retail since 2004. He founded Internetretailing and RetailX. This week he had not one, but two great pieces on Agentic commerce on his substack called ‘In No Particular Order’. Together they are a complimentary set of pieces that are a great reminder for agent builders to remember, you have to put the customer first.
In Part II, he had several readers of Part I ask for an example of the ‘Customer Sovereign Agent’, so he goes through an example with a personal angle:
How Target is Using AI
Similar to Walmart’s WSJ announcement of 4 ‘super agents) (covered in Retailgentic’s Week 30 update)
Their Super Agents are called Store Companion (AI chatbot for store staff) and Precision Plus (Optimization AI for their Retail Media Network called Roundel. Details in Digital Commerce 360 here.
Agentic Commerce Goes Mainstream
Prior to this week, Agentic Commerce the biggest mainstream publication to cover the Agentic Commerce trend was WSJ and Barrons and that was covering specific companies like Walmart and the news sites.
This week we had a plethora of mainstream pieces come out. These aren’t ‘great reads’ but do a good job of summarizing what we’ve been tracking for a while and introducing many of the concepts to a broader business and consumer audience:
CNBC
This CNBC piece goes over the ChatGPT rumors ahd the seismic shift it is set to cause in internet shopping.
Entrepreneur
Entrepreneur magazine has a piece targeted towards SMBs and service businesses. It explains the overall traffic trends away from organic search and to how GenAI can counter the effect
New York Magazine
Like Entrepreneur, this NY Mag piece, has less of a consumer orientation, but also chooses to introduce the concept of Agentic Shopping via GEO and how it ‘tricks AI chatbots’.
Pre-GenAI Legacy SaaS Ecommerce Vendors Chime in
This is a bit ‘inside baseball’, but this week a slew of traditional pre-GenAI legacy SaaS e-commerce vendors realized that ‘Agentic Commerce’ is a thing and took dramatic action and…wrote some blog posts 😁. I don’t know why this week was ‘the week’, perhaps it took ChatGPT-5 to make them realize this was not a fad.
Seriously though, you’ll see all of the traditional SaaS vendors who are now used to 3-5yr cycles have to adjust to the market changing every 3-6 months. It’s going to be interesting to see which can make the pivot and some hard choices that come along with that, or fail to pivot and potentially end up in an innovator’s dilemma.
Here’s a roundup of four I saw:
Rithum (prev ChannelAdvisor+CommerceHub)- Agentic AI is skipping your products here. In addition to a blog post, they also changed their brand and website around a bit.
BigCommerce - To show how serious they are about Agentic, they pulled two levers:
First, they changed their stock ticker to from BIGC to CMRC for ‘Commerce’ (I think ECOM is cooler, but, hey, I’m slightly biased).
Second, they announced a significant rebrand - Now they are a combination of BigCommerce, Commerce, Feedonomics and Makeswift. Rick Watson, in his always consistent, no punches pulled style, has a ‘Future of Commerce’ hot take on LinkedIn here.
Mirakl - Not to be outdone, Mirakl announced that their previous collection of third-party marketplace sellers, called Nexus is… you guessed it - now an Agentic Commerce Brain 🧠 .
Channelengine - This one’s a surprise: Norway-based EU ChannelAdvisor competitor, Channelengine, is….DANG you guessed it AGAIN!, embracing Agentic Commerce.
Coming up this Week
On the Retailgentic pod, after a brief break this week to give our production team a breather, next week we are back with the first of a killer line-up of guests. Thematically, they are all going to help us navigate where Agentic Commerce is going to so that, together, we can triangulate on where where the puck is going. Stay tuned!