When we announced this year’s BRXND conference, we said that “there is a not-hard-to-imagine future where the most important job of a marketer is to convince the model it is worthy of its user's attention and dollars, and, even further, where the LLM just has its own credit card and makes the decision on a consumer's behalf.”
That future has arrived. Today’s post is a quick reaction to Meta’s launch of Muse and the quantum leap forward in agentic commerce that has taken place in the past few days.
The discussion continues on stage at the Times Center on Nov. 5 with outstanding leaders from Amazon, Microsoft, Bark, Rocket Money and more taking the stage. Can’t wait to see you there.
Muse & The Agentic Commerce Conundrum
Agentic commerce is suddenly very real folks.
Two weeks after Instinct initially took the tech world by storm, the hills of the Twittersphere are alive with the sound of people executing complex purchases end to end with a few simple texts in iMessage. Almost overnight, we’re seeing the inklings of a true agentic shopping experience that will really feel superior to many shoppers and not just on commodity, emotionless purchases.
Now, as he so often does, Zuck has come in off the top rope as a fast follower with a personal agent called Muse that will quickly bring this technology to the masses. Early reviews suggest that Muse is very, very good across both life admin and agentic commerce use cases. A month ago I asked when consumer AI would have its “Claude Code moment.” This is it. Yet again, the road to financial run is paved with people who bet against Meta.
With each passing day, it feels increasingly likely that we’ll work primarily within harnesses in the next 12-24 months. Now, if Meta can shape the future in their image and likeness, we’ll run our personal lives and shop in them as well.
There are a plethora of fascinating second order questions to spill ink over here but top of mind for me is how brands of all stripes respond to this sudden influx of consumer agents. We’ve already seen the first flashes of tension with Resy deactivating user accounts after Instinct made 200+ requests in an hour. This is a perfectly reasonable response and microcosmic of a much larger point of tension.
Simply put, the best agentic commerce experience for users is going to often be at odds with optimizing yield for merchants. Tools like Instinct, Muse and Grok are going to mercilessly hunt for promo codes and off-menu discounts and will consistently try to “beat the system” in unforeseeable ways on behalf of their users. But beyond that, what really is going to make many of the largest incumbent platforms on the internet viscerally uncomfortable is that they no longer control the shape of a buyer’s purchase journey.
Agentic commerce breaks many of the meticulously optimized purchase funnels (often at razor thin margins) that consumer businesses have spent the last two decades building online. It’s also incredibly unclear if or how entities like Muse and Instinct will be influenced by traditional sponsored product ads as they decide what to purchase. This is going to cause a major conundrum for merchants, especially because from a profit perspective, most large retailers are now advertising businesses masquerading as purveyors of goods.
Brands are understandably skeptical of the version of the future in which they get disintermediated from their customers. There’s not much nobility in being a wholesaler and logistics arm…and you never beat Amazon at that game anyway. I’ve long posited that many have effectively been playing a version of narrative arbitrage where they are enthusiastic about agentic commerce in public to appease the street but behind closed doors, they are happy to see early pilots struggle to gain traction. Walmart enthusiastically backed OpenAI’s early embedded checkout experiments but was pretty quick to publicly share how much lower the average yield was off their own platforms. That “agentic commerce converted 3X worse” number sure spread in the press like wildfire.
That said, I’m far less convinced then Eric Seufert that open agentic commerce outside of retailer owned ecosystems is entirely a “mirage”. Blocking off your marketplace writ large to an agent eager to quickly buy something from you (and likely operating on behalf of a person who spends a lot more money annually than the median shopper) is cutting off your nose to spite your face.
It’s far better to be early to a platform shift that is happening gradually than late to one when it happens suddenly.
The Three Miracles of Agentic Commerce
Historically, the macro knock on agentic commerce has been that it further solves buying but kills shopping in the process.
Last month, I put forth a framework that inspired lively discussion on LinkedIn around the “three miracles” that need to happen for agentic commerce to ultimately be better at helping users discover what to buy than the status quo experience in online commerce today.
1) Merchants must expose enough structured data and tell a consistent brand story at both at the catalog level and across the web for agents to understand how individual products match to highly semantic and fuzzy intent.
2) When items are sold at multiple retailers (or sometimes when choosing what to recommend in first place), agents must consider which merchant provides the best overall value as a function of price, loyalty status, rewards, delivery window, and more.
3) Agents must possess enough “media literacy” to understand which data sources are reliable arbiters of product quality. In other words, they must discern pay to play slop masquerading as organic product reviews from publisher or creator content where an objective reviewer actually tests a product.
While #2 and #3 are largely in the domain of the companies building agents, point #1 is directly on brands and by far the most impactful lever marketers have to control their destiny in agentic commerce, at least until we can all buy ads to influence agents at the point of inference.
While there’s much handwringing from the tech Twitterati about if agents should be broadly anthropomorphized, I strongly believe that agents will effectively be more extreme versions of human buyers when it comes to gravitating towards brands that are story rich and consistent with their messaging across all key touchpoints. As Noah wrote back in December:
”When you RLHF a model to be a “helpful assistant,” you’re essentially training it to care about context, explanation, and story (emphasis mine)—exactly the things we thought the robot would skip past.”
What I’m really trying to say here is that however big a seat your company thinks the marketing team should have at the agentic commerce table, make it bigger.
Shopfiy’s Field CTO Rohit Nathany put this cleanly: “I highly recommend my marketing and growth friends start treating catalog and product data as a marketing tactic, not just an admin task. Your catalog should be complete, structured, and story-rich.”
On that note, I’d love to jam with any marketers who are currently working on this problem— drop me a line at mike@brxnd.ai.
The Brand Brain & Turning Context into Competitive Advantage
As more work moves into AI tools, a company’s advantage comes from how effectively those tools can use its own knowledge and expertise. Said another way, context is king and a “brand brain” is how companies turn that context into an enduring competitive moat.
With assistance from Youtube maestro Sam Knight, Noah has been on fire lately on the video and podcast circuit and we’ll be bringing much more of Noah and Sam’s work to you here.
In the video below, Noah unpacks what a Brand Brain looks like in practice: connecting a company’s data, context, expertise and tools so AI can work with the knowledge that actually makes the organization distinctive.
Geno Smith vs. Armageddon
Football is back….which means I once again have to remind myself that suffering through another New York Jets season is a choice I freely make. Nobody is forcing me to do this! I can pull the plug at anytime if this isn’t even achieving the goals I want it to.
Anyway, if like me you’re mildly distressed that an Anthropic staffer seems to genuinely believe that there is a 10% chance AI will cause mass extinction and he’s powerless to stop it, here’s a quick post to put your mind at ease.
I would at least like to see one 34-13 wild card round loss to the Baltimore Ravens on a miserable sleet-soaked January afternoon at MetLife before AI brings upon my imminent demise. Is that really too much to ask?
If you have any questions, please be in touch. As always, thanks for reading.
— Mike
from the brand side: we spent years making brands legible to people, and now the system has to be legible to a model that never sees the moodboard. written-down taste - naming, voice, structure - is what an agent can actually read. the brands that documented theirs will be the easiest to recommend ✨