If you are sceptical about the advent of agentic commerce, don’t be. So much of day-to-day commerce is boring, and so much of what is not boring is baffling. In either case, I’m looking forward to handing it over to the bots.
Agentic Commerce Calculations
I got an email from one of the hotel chains that I use. I have reward status with them, so I opened it and read it. They were offering me bonus points for buying points for cash. I travel a lot and stay in hotels a lot, so maybe it would make sense for me to do this. But how to work out whether to buy or not? It was straightforward. All I had to do was open up multiple windows with hotels in the chain in each of the cities I might be likely to visit over the next year then look at the price of the rooms for the approximate dates that I would be visiting. Then I needed to look at the price of the same rooms in points and work out the average points exchange rate. From there, I just made a graph comparing the bonus points offer (there were relatively more bonus points for buying more points to begin with) to look at where the bonus point cost fell below that exchange rate.
Of course I didn’t. I had other things to do.
Similarly, when I last needed to book some flights for a family event, I had an airline companion voucher and an upgrade for two voucher. So to work out whether to pay for the flight in cash or to use one of the vouchers, I looked at direct and one stop options for the flight on a spread of days and then made a spreadsheet with the money cost, the points cost, the money plus upgrade cost and the points plus companion cost. This allowed me to calculate the dynamic exchange rate between points and money so I was able to determine that it was better to use pay cash and use the points for an upgrade rather than pay cash or use either of the vouchers.
Of course I didn’t. I had other things to do.
Last month, like a great many other Brits, I had to work out whether to add a contribution to one of our tax-free savings accounts (these are called Investment Savings Accounts, or ISAs, in the UK and there is a limit to how much you can deposit in these each year) or whether to open a new one. So I looked up a table of the accounts paying highest interest and then made a shortlist noting the terms and conditions (some accounts allowed me to transfer in from other accounts, some didn’t, and so on). Then I opened up a new account by showing my passport and moving my head around and downloading a PDF of a utility bill and then uploading it.
Of course I didn’t. I had other things to do.
Bots don’t have other things to do though, so a bot could easily have told me that it would be a good deal to spend $500 on points and to ignore the offer the buy frequent flier miles and to open a particular new account and transfer the money from the other accounts. In fact my bot could have negotiated with the hotel bot to get a better rate in order to get me to favour that chain for my stays over the coming period and done a deal with the airline bot to book flights where it could auto-upgrade me using miles and connected with the financial institutions bots to automatically .
(Compare this vision of the future to what actually happened recently when I changed my schedule. I needed to book a night in a hotel chain but I got the password wrong when logging in on my phone at the airport — they weren’t smart enought to use passkeys — so I booked a night in a different hotel chain where I could remember the password. All that marketing money, wasted.)
So what will it mean for hotels, for financial institutions and for airlines in this new world of agentic commerce? I think it is clear that, just as the transition of e-commerce meant new infrastructure, the agentic shopping future will require implementation of new technology, new capabilities and new infrastructure. And as companies such Walmart already recognise, it will mean the service providers have to build their own agents, because they will want to implement AI-AI negotiation and not be reduced to pipes while AIs use their web sites and APIs.
When you think about the implications, they are significant and will mean an existential threat to some value chains. We are looking at a paradigm shift in the truest sense of the word. Ken Mandel summarises the nature of the shift in advertising terms rather than nicely. He says that today humans see and interact with advertisements, but in the near future the consumers will be using agents, so they will never see the messages targetted at them. In the not too distant future, though, the brand agents and the consumer agents will interest directly through structured data sets. In other words, it is time to “think code, not copy”. I know nothing about the advertising business, but this perspective seems both valid and a a template for other sectors as B2C and B2B2C models are steadily replaced by B2R2C (business-to-robot-to-consumer) models.
What Does Agentic Commerce Change? Everything
What this all means is difficult to predict. Whether it is hotel points, or frequent flyer miles or interest rates or whatever else, the fact is that advent of agentic commerce will change the relationship between providers and customers and therefore the business models for the market participants. And this is where I am focusing my research work right now, because I genuinely do not know what will happen to those business models as each sector moves from a B2C to a B2R2C model, when all consumer deposits go to accounts that earn the highest interest and when all frequent flier miles are redeemed.
However, I think I can make the strategic prediction that banks, hotel chains, retailers and everyone else will need to start building agents to talk to the customers’ agents, because the customers will have other things to do.