I think we’re getting to a much more useful stage of AI. These systems are moving beyond simple chatbots and becoming working companions that can actually do things on our behalf.
That’s the part I’m most interested in. How does this help us be more efficient in our personal lives, our working lives and our businesses? A better answer to a question is useful. Having the system carry out the task opens up a much bigger set of possibilities.
One of those possibilities is shopping. If an agent can understand what you need, compare the options and complete a purchase with your approval, the way people interact with businesses starts to change.
Think about the ordinary things it could do
Imagine you’re in the kitchen, looking through the fridge and working out what needs to be ordered. You tell the system what you’re missing, and it helps put the order together. You don’t have to stop what you’re doing, open another app and start the whole process yourself.
Or think about booking a trip. There are dates, prices, hotels, travel times and preferences to work through. An agent that remembers the relevant details could do much of that work before coming back with something for you to approve.
These are fairly ordinary examples, which is why I find them interesting. The technology becomes much more useful when it fits into something people already do every week.
Some of that capability is already here. OpenAI’s agent announcement in July 2025, for example, described a system that could navigate websites, help plan grocery purchases and ask for permission before taking consequential actions. There’s still work to do, but the direction is clear.
I wouldn’t be surprised if, over the next couple of years, interacting with these systems becomes much more natural. Less opening a chat window and thinking of a prompt. More being able to continue a task whenever an idea comes to mind, with the system remembering enough context to be useful.
That’s my expectation, rather than something I think will arrive everywhere at once. People will have different levels of comfort with it, and the technology still has to earn their trust.
More tasks could become transactions
Once you start thinking about agents doing things for people and businesses, you can see how this goes beyond groceries and travel. A task might involve purchasing software, paying for data or getting access to some other digital service.
There could be a lot more individual transactions happening on behalf of a person or a company. The person gives the direction, and the system carries out several steps that previously would have taken someone’s time.
Keep in mind, more agents doesn’t automatically mean people have more money to spend. Some transactions would be things they were going to buy anyway. Others might become worthwhile because the time and effort involved have come down. It’s too early to know how large that effect becomes.
I also don’t think it’s sensible to talk about this as though there won’t be complications. If a system is spending on someone’s behalf, it needs clear permission and limits. Getting a product recommendation wrong is one thing. Acting on it with someone else’s money makes the consequences more immediate.
There will be mistakes, and I think we’ll go through periods of excitement and disappointment as people work out what these systems can reliably do.
But I’m excited about the overall direction. If you’re good at giving clear instructions and judging the work that comes back, that becomes a useful skill with this technology. More of the time can go into deciding what needs to happen, with less spent carrying out every step yourself.
That’s a meaningful change for a business owner or a busy professional. I’m interested in how quickly it becomes part of ordinary life.
