AI agents in digital commerce might be new, but awareness is already high. OpenAI not only promoted its Operator feature and doubled down on it by providing additional tools for developers soon afterwards, but also launched it in concert with a number of brands enabling customers to streamline their online shopping with AI agents.
OpenAI said they were collaborating with “DoorDash, Instacart, OpenTable, Priceline, StubHub, Thumbtack, Uber, and others” which gives a good indication that a range of industries see AI agents as an opportunity. With Amazon integrating the “Buy for Me” agentic option into their platform as well, it seems clear that agentic activity is going to touch every industry.
At Forter we’re seeing that range reflected in the kinds of purchases customers are inclined to use AI agents for. At the same time, there are some early preferences emerging in the kinds of purchases where customers more naturally lean in to delegating authority to an agent.
Groceries Are a Go-to
Consumers love to pass their grocery shopping onto AI agents! Forter’s data shows that so far, customers are 2x more likely to delegate authority to an agent for grocery purchases than for any other kind of purchase.
I didn’t expect to see this when I started looking into the data, but it makes sense, because grocery shopping ticks a lot of boxes for the kinds of cases where I can see myself using an agent. For example:
- Weekly shop: When it comes to groceries, there are so many things that I order every week, or every other week, or once a month. I already have saved common items. Why not just hand the whole thing over to AI?
- Specific recipes: If I’m trying out a new recipe, I’ll pull the ingredients from there into a shopping list. There’s nothing in that process that requires me, and honestly it’s pretty tedious. I’ll definitely try handing this over to an agent. A lot of the time the recipe itself is online to start with, and sometimes it even comes from ChatGPT, so with an agent in the loop as well the entire process should be exceptionally smooth.
- Planning a party: With party planning, I’ll often start out by throwing everything I need to remember or prepare or buy into a doc or spreadsheet, to help me order things in my head and so I don’t forget things later. Now, that also translates to making it easy to delegate authority to an agent for the actual purchases.
The Speed is the Thing
Groceries aren’t the only area where customers find it natural to try out an agent as part of the purchase process. They’re also exploring delegating authority in any case where speed is a critical factor. In fact, it’s 1.5x more common to see agentic activity with these cases than in purchases where speed is not relevant.
That happens more often than you might think. For example:
- Concert tickets: Tickets for popular concerts can sell out fast, leading to disappointment if you aren’t able to be at the checkout at the right moment. It’s much easier to hand it over to AI.
- Sporting events: Once you know that your team will be playing in a particular match, whether it’s an important game or one that you especially want to go to, then time may be of the essence in making sure you get the ticket you want.
- Deals: We all love a good deal, but we can’t always grab it at the right time. Whether we’re busy at the wrong time, or simply miss the opportunity, it’s always frustrating to find out that it’s too late. Consumers who delegate authority to an agent to catch the deal when it’s live, or snap up an item when it reaches a specific price range, don’t have this problem.
- Limited-drop items: Some goods can only be purchased in a small time window. If you really don’t want to miss out, then relying on an agent for the purchase might be easier than handling this yourself.
Without an AI agent in the picture, these sorts of purchases can be quite stressful, and might need to take place at inconvenient times. With an AI agent, all the hassle is gone.
Delegating Authority Thoughtfully
What these trends indicate to me is that the early adopters of agentic technology are using it thoughtfully. They’re delegating authority to agents in cases where human intelligence is not needed, and artificial intelligence can make the process smoother, faster and less stressful.
Consumers appear to be happy to keep the creative side of things in their own hands. Finding and choosing new recipes, for instance, or deciding how to prioritize hobbies and spend leisure time, or working out what price range feels right for an item they want.
Once the decisions have been made, there’s no reason not to involve an agent. The thinking has been done, so why waste precious time implementing the details?
What’s Next?
I don’t have a crystal ball (although I could get an agent to order one for me) and I have no idea what lies ahead with this technology. What I do know is that with so many large brands and sites finding ways to incorporate agents into their purchase flow, consumers are going to have a wide range of options for how they use them.
So far, it seems that people are willing to give it a try. It’s too early to tell if the use will expand beyond early adopters, but to me the fact that after only two months the most obvious and intuitive use cases are already dominant suggests that this technology has a natural home in the e-commerce ecosystem.
It’s a huge opportunity for merchants to become embedded in customers’ regular, automated shopping habits. It comes with challenges, both technical and fraud-related, but OpenAI is trying to make it technically easy and as I’ve written before, the fraud challenges are more easily solved than you’d think. The bar for successful merchant adoption is relatively low.
Shopping online already saves me a huge amount of time. Using an agent can save me a bit more time, and more importantly can save me hassle as well. Given how stressful life can be, I think I’ll be giving that a try.