A case for agentic payments

One type of agentic usage we are still being promised to see is agentic payments. Right now, the only people really excited for this are agentic payment startups. The core tension is that who out here wants to spend MORE on online shopping?

Besides, the idea of giving your wallet to an LLM feels creepy, something that only a deranged give-your-live-to-openclaw AI-psycosis person would do. Nevertheless, I do see one scenario where chatting with an agent and then it asking to confirm a purchase, feels interesting.

My case is built upon another theory that I still have to lay the foundations of, but online shopping apps are annoying at best. I am sure lots of very talented people work on them, but I always use them out of necessity (I can't find what I want on physical stores) or because it's cheaper. I don't see online shopping as a pleasurable experience.

For example, I want to buy a new phone. I have an idea of what I want, but all shopping filters are unreliable, or hard to set up: I need to navigate multiple hamburger menus, check here, drag there, type something, so I pin down to what I am looking for. I press search and 2 out of 5 results are off, because everything is a marketplace and it relies on unreliable input of merchants. Then, there's the ads, the new feature they added that I don't care, the popups...

AIs can dig through this mess much better than me, and since it feels like a waste of time when I do it, I do think there's a case that an agentic shop experience can reduce frustration and the time it takes from opening the app and closing the checkout. At the same time, this case also implies that the user is not going through all the A/B tested machinery to keep you spending there. Is it worth it? Could agentic payments reduce overall conversion or the shopping cart size because the user wouldn't interact with it directly?

This is all very speculative, admittedly.