AI agents are coming for the inconvenience premium

A business that keeps customers because leaving is exhausting should treat personal AI agents as a threat to its margins. The most interesting thing about an assistant that can negotiate a bill is what happens to the company on the other end.

On the 24 September edition of The AI Daily Brief, Nathaniel Whittemore explored the argument over whether consumers actually want agents doing things for them. Shopping can be enjoyable. Booking an appointment or completing another form may be less so. That distinction matters. But for business leaders, the sharper distinction is between customers who choose to stay and customers who cannot be bothered leaving.

Both renew. Only one is loyalty.

Meta says its new personal agent, Muse, can fill out forms, negotiate on a user's behalf and keep working after the app closes, checking for approval before making purchases. At Connect, it announced plans to bring Muse to its glasses and introduced a pocket device called Charm. Glasses integration was promised for the coming months; Charm came with a promise of more information later this year. These are announcements, not evidence that everyone is about to delegate their shopping.

The hardware is interesting because it could make delegation less of an event. See a renewal notice, give an instruction, get on with dinner. Whether the instruction travels through glasses or a phone matters less than whether the customer has to finish the job personally.

There is an established business problem underneath the gadget story. In its 2022 evidence review, Britain's Competition and Markets Authority described how excessive friction can obstruct consumer choice. Its examples include services that are easy to join online but require an offline process to cancel. The review explains how the effort involved can cause customers to abandon cancellation.

That gives a business an uncomfortable possibility to investigate: some of its apparent pricing power may be an inconvenience premium. Customers might accept a worse deal because fixing it consumes more attention than they want to spend.

Consider a deliberately simple example. Saving $15 a month means $180 a year. If getting the saving takes an hour of forms and follow-up, a busy customer might leave it alone. If an agent could complete the same task reliably for $25, including the cost of checking its work, the decision would look different. Those are illustrative numbers, not a claim about Muse's pricing or performance. The point is the threshold: cheaper effort makes smaller grievances worth pursuing.

That could be awkward for a company whose retention strategy depends on customers eventually getting tired. An agent need not be brilliant at everything to put pressure on that strategy. It needs to be useful at the particular chore standing between a customer and a better offer.

There are limits. Meta itself acknowledges that Muse will make mistakes. Its launch announcement described a US rollout, which should not be confused with Australian availability. Neither the launch nor Connect establishes durable consumer adoption or the economics of completing a disputed cancellation. A system that needs its owner to supervise every step may simply move the paperwork into a chat window.

Nor should we assume the new intermediary will be neutral. Meta's Connect announcement lists retailers and payment providers it is connecting to Muse. Partnerships can make transactions easier, but they do not demonstrate that an agent will search every available supplier. Business leaders should watch whose offers get considered and on what terms. A customer can escape one source of friction and still become dependent on someone else's selection process.

For an Australian business, the useful response is to examine one renewal journey now. Ask a staff member unfamiliar with the process to compare an offer, request a change and leave. Record where they need help. Separate checks that protect the customer from steps that merely discourage them. There is a material difference between verifying someone's authority to close an account and making them explain their decision for the fourth time.

Then look at the commercial exposure. If customers could compare the full cost and switch with much less effort, which revenue would be vulnerable? Treat that as a planning scenario, not a forecast. It is a better starting point than counting how many staff have tried a chatbot.

The opportunity is just as practical. A smaller supplier with a competitive offer should make its terms clear and the purchase straightforward for a customer or their authorised representative. Being easy to evaluate could become a sales advantage. Paying for more persuasion while leaving basic terms obscure would be a strange response to buyers acquiring better help.

Personal agents may disappoint. Wearable hardware may struggle. Neither possibility makes customer exhaustion a sound competitive advantage. Start earning the renewal you would receive if leaving were easy.