It seems only a minute ago that we were all waking up to this brand new thing called generative AI, and trying to figure out how we could make best use of it for our businesses. But even as we tinkered and tested, AI kept evolving before our very eyes, forcing us to keep recalibrating our ideas.
And so the process goes on. Generative AI has already had time to morph from a content-making engine into search engine. But this year, two new feature additions in ChatGPT perhaps sum up best where we’re heading next.
First, in July, OpenAI launched Agent Mode in ChatGPT. While the standard chat interface responds to questions and prompts by generating outputs, in Agent Mode, ChatGPT can handle entire tasks and workflows independently, including executing actions in other platforms.
Then, in September, ChatGPT launched Instant Checkout in the US. Initially operating in partnership with Etsy, it lets customers search and pay for products right there in the ChatGPT interface. Pair it with Agent Mode, and you can even give the AI a brief for what you want to buy, your budget etc. – and then leave it to the machine to make the purchase decisions for you.
This is just one example of what has already been dubbed agentic commerce – the use of AI tools as automated shopping assistants with the power to buy on your behalf. One in 10 Americans have already used agentic AI to make purchases, mainly for small-ticket grocery purchases. But that number is expected to grow rapidly, with a third of Gen Z consumers saying they are keen to try it.
So far, most of the discussion around agentic commerce has focused on its implications for ecommerce, especially around what it means to make web stores ‘machine-readable’ so AI rather than human consumers can engage with them. But with less and less separation between digital and physical retail, it would be short-sighted to think that agentic AI won’t have any impact on in-store operations.
So the question is, what might that impact look like?
Agentic AI as Digital Broker
To start with, it should be acknowledged that any role agentic has to play in physical shopping journeys will be qualitatively different to online. The big appeal of using an AI agent to shop for you online is that it takes full responsibility and automates the whole process. If you’ve headed out to a store to have a look around yourself, you’re clearly not looking for that. But that doesn’t mean an AI agent can’t be a useful assistant.
One area where it’s easy to imagine AI agents playing a role is payments. A recent article in Forbes covered the emerging trend for agentic AI to play what amounts to a broker’s role in arranging instant credit at checkout, for example by comparing interest-free, short-term credit arrangements offered by BNPL operators like Klarna and Affirm with the longer-term payment plans from their credit card provider.
Again, the article focused on ecommerce scenarios. But the same could play out in stores, too, with retailers running AI agents on their POS system and self-checkout kiosks to help customers find the best credit arrangement for larger value purchases.
That would see AI agents run on the retailer side for customer use. But what about customers coming into the store with an AI agent they use as a personal shopping assistant on their cell phone? How can stores cater for that?
That will depend on facilitating connections between a customer’s personal agent and your in-store digital systems. You will need an agent-ready infrastructure with the necessary APIs and over-the-air connectivity to let a visiting AI agent ‘talk’ to your systems.
The customer’s AI knows all about their preferences, habits, purchase history etc. The idea would be to run a matching AI over your systems to provide what would amount to a machine-to-machine clienteling service, learning about the customer through their agent, and then making recommendations or pushing certain offers and promotions based on that information. This opens up an effective route to hyper-personalizing the in-store experience, and could even pave the way in the future to dynamic pricing based in part on AI-to-AI negotiation.