Marcos Angelides admires the autumnal fashion stylings of Carmen Berzatto, the titular character of HBO’s The Bear. Naturally, the co-managing director at Spark Foundry fed one of Carmy’s jumpers into ChatGPT and asked, “I don’t know how to style it, can you help me out?”
The bot quickly gave Angelides suggestions for how to style the pullover, and which social occasions and levels of formality it would be suited to.
Wanting something more visual, the agency MD guided the AI to show him what this might look like. ChatGPT asked about his appearance and then customized the results accordingly.
“Suddenly I’ve got my own personalized lookbook,” said Angelides. “Everyone’s effectively now got a personalized stylist. It’s really, really good.”
He gave a top tip to the audience at this month’s Festival of Marketing (October 3), where he was speaking about AI personas and the future of commerce: “Get some of your clothes, put them all on your bed, take a photo and ask [ChatGPT] to organize all your outfits for the week.”
But back to a Chicago sandwich shop in the fall, and the hunt for the right togs – Angelides asked the AI, ‘Where can I get these outfits from?’ and ChatGPT listed out retailers and specific products with different price points, pros and cons, and links to buy.
“The only thing that’s missing is me giving it the credit card and saying, ‘make the purchase, I want to spend no more than a hundred dollars… I need it by Friday because I’m going out on the weekend,” said Angelides.
AI in commerce is “going to change the entire way that we shop online,” contends the Spark Foundry co-MD.
Angelides was talking at the Festival about the rise of AI personas; from creators like Caryn Marjorie ‘AI-ifying’ themselves, to Meta’s AI Studio and TikTok’s Symphony and the potential for a virtuous circle where an influencer’s ‘back catalogue’ of content becomes valuable for training an AI and therefore incentivises them to post more often and in doing so creates ad dollars for the platforms in question.
What does this have to do with commerce? Well, aside from the arguable potential for AI personas to drive live commerce day and night in the language of your choice, Angelides makes the more general point that people are increasingly open to interacting with an AI, whether that be a chat bot or a full-on uncanny-valley character.
“A lot of the conversation right now [in the enterprise] is very focused on generative AI as a business tool, as an efficiency driver,” said Angelides.
But beyond making things cheaper and quicker, “all of which is very important,” we’re “ignoring the most important aspect, which is… the role of AI in culture,” he added.
“How is it reshaping people’s behaviors, the relationships they have with each other, and ultimately the way they behave online and their expectations for brands?”
Angelides declared his interest in GenAI as an “effectiveness driver”, rather than just an “efficiency driver”.
At Econsultancy we’ve covered AI-powered commerce search, what GenAI and RAG systems could mean for the shopping experience, and how some retailers have been experimenting with UX that moves away from faceted navigation and towards social-style homepages or chatbot-guided discovery.
The contention is that mission-driven queries are on the rise, and in some contexts shoppers want to be inspired or receive recommendations without having to deal with faceted navigation from the get go.
Search in grocery retail or marketplaces with their thousands of SKUs is typically the business case that is being explored, and Angelides concurs this is ripe for disruption.
“It takes forever to do a weekly shop. If you do it from scratch, it’s about 60, 65 clicks… It’s still a point-and-click solution. You have a main navigation, you click on it, you go to sub navigation, click again, you click again…”
Angelides is bullish, saying, “Very soon we won’t click at all. it’ll be zero-click commerce, we’ll just have a conversation with the AI and [ask it to] write a weekly food plan for a family of four – the kids have certain eating requirements, I’m not very good at cooking, so I need help organizing all the ingredients, put together some recipe cards for me, I don’t want to cook for more than 20 minutes, on Sunday there’s 10 of us so help me do something really, really impressive, and create images as well so I know exactly what it looks like. I work at home on Thursdays, make sure it’s delivered before eleven o’clock.”
Amazon’s Rufus chatbot was namechecked by Angelides. It finally rolled out in the UK this month and the Spark Foundry MD described it as “game-changing”.
“Amazon is great and convenient when you’re shopping for low-ticket items, but it becomes a very, very lonely place the second you want to buy something expensive. You’re suddenly very aware, you’ve got no expertise, no guidance, you’ve got to try to read through everything, and if you’ve got quite a specific question it’s impossible to get an answer. That’s where Rufus comes in.”
Angelides says this advice from a chatbot makes people more confident to buy and reduces returns because expectations are managed. He describes Amazon as having “fired the starting gun for all other retailers in terms of how they need to rebuild their architecture”.
What are the implications of a chatbot or an AI search experience recommending a product, as in Angelides’ lookbook anecdote?
“I didn’t ask it to select those retailers,” he said. “How do you make sure you start turning up in these search results? What does it mean for branding in a world where you’re dependent on AI to shortlist?”
Angelides says that this discussion shouldn’t be approached purely from a “performance perspective”. “This is about branding, in order… for the AI to think you are completely the best brand for me to recommend to my user,” he added.
And in the case of building a grocery shopping basket, brand plays its part here, just as it does in brick-and-mortar stores, where customers will refuse to compromise on certain brand products. “You might say that for certain categories of things, you just don’t care what you get. But [for] something like chocolate, you may have a very specific brand [in mind], and when it comes to meat, you know you want it to be organic,” said Angelides.
The idea of mental availability, already so important in marketing, serves as a “way to hack and get round the AI agents”, as Angelides puts it. No doubt retail media also grows in importance in this imagined future.
Angelides advises brands to understand the sorts of “long-tail conversations” people are having – the rich context around what people are shopping for. He adds that, effectively, product discovery and UX is moving from “a first party data solution into a zero party [one]”, where retailers understand “real context about your life and your needs, rather than just, what have you shopped before and how have you landed on a website”.
Some comfort might be found in Angelides’ takeaway about ‘AI SEO’, which is that it “reinforces a lot of the fundamentals of marketing”.
“It’s going to be based on things like… top SEO rankings, what is getting social media buzz [or PR coverage]. The drivers in marketing won’t change, the mechanics will…”
“Auditing is the easiest part to start doing,” says Angelides, saying that brands should be asking AI chatbots for recommendations in their categories and seeing where and how they currently appear.
“That’s the easiest first step and then you can start making a bit of an AI worry map,” he adds, advocating that marketers start to work with their agency partners on anticipating how commerce behaviours may be about to change.
JD Sports on AI search and digital transformation: we can add ‘so much scale to personalisation’