Why AI agents are coming between companies and customers as tech risks consequences for internet economy

The internet’s future may see people no longer visiting individual apps or websites, instead relying on AI agents to do most of the browsing for them.

The shift being plotted in Silicon Valley would erase the internet playbook of the past 20 years, where companies built websites to attract human visitors, persuade them to buy products, read news, book travel, compare prices, or arrange appointments.

But intelligent agents are now learning to do those jobs for their human owners, with US tech giants racing to grab market share in a battle to control the next-generation digital economy.

On Monday, shares in Instagram-owner Meta surged 11 per cent as its AI agent named Muse won strong reviews and rapidly gained users in the US.

Mark Zuckerberg’s Muse agent will compete with similar products from other tech giants in what could in what could quickly become a market serving billions of users each day.

“This will be a bigger battle than anyone anticipates,” said Nikesh Arora the chief executive of $US304 billion cybersecurity giant Palo Alto Networks.

“It’s only a matter of time before there is an Apple and Google version of Muse and possibly TikTok, in addition to the frontier LLM agents [Anthropic and OpenAI].”

The emerging shift would put a new layer between businesses and their customers, giving the companies controlling AI agents huge power over which products, information or services get chosen.

On September 19, Amazon responded by saying it would ban Meta’s Muse agent from scouring its website to order goods on behalf of its human masters.

Amazon says Muse is unauthorised and is likely worried its website’s traffic will plunge if humans no longer need to browse websites themselves.

Calls for regulation, slowdown

The risk of disruption from AI and its agents prompted calls for more regulation and warnings about the tech’s risks last week.

But Wall Street’s surge to record highs on Monday suggests investors don’t expect the tech’s progression to be limited.

For news publishers and retailers that rely on visitors to their websites to generate subscriptions, advertising, and sales, the threat from AI agents is particularly acute.

In Australia, news publishers are demanding the AI giants compensate them for the content they’re arguably ripping off for their own commercial benefit.

Amazon has simply attempted to ban the agents from visiting its website.

The shift in the internet economy and capital allocation means a lot will need to be negotiated among companies, regulators, and governments on a cross-border basis.

Security, privacy, financial fraud, governance, copyright and competition law abuses remain at the top of AI critics’ agendas.

Fast-moving shift

Much of the race to invest trillions in AI is ultimately about the internet’s landlords — the likes of Meta, Microsoft and Google — strengthening their stranglehold on society and the digital economy.

The political backlash against data centres and mega-cap tech’s dominance in the US has gathered momentum in 2026, while Prime Minister Albanese laid out a national framework for, AI in Australia’s Interests, in a Sydney speech in July.

Analysts though still expect Silicon Valley to grow its global power thanks to the rise of AI agents working on behalf of humans.

“The most dominant AI assistants will probably come from Apple, Google, Amazon, ByteDance, or Meta, but they’ll want to [and] need to make money from them,” says Dennis Hong the Chief Investment Officer at ShawSpring Partners.

“The easiest way is to charge businesses. A hotel could pay to be suggested first [by the agent]. A restaurant could pay a fee on every order. You might think you’re getting the best option when you’re really getting the one that paid.

“Apple and Google likely have an advantage with consumer because they make the phones and can see your calendar, messages, and habits. Governments especially [the] European Union will probably step in to regulate this activity.”

The agents are also emerging as an effective replacement for Google Search and mean companies will potentially need to reach customers or readers via agents, rather than search engines.

The businesses and management teams behind popular mobile apps as gateways to their internet products, or services, will also need to decide whether to connect to the agents, or not.

Backed by Wall Street’s financial firepower, Zuckerberg and Silicon Valley’s tech oligarchs will be betting their monopoly-like networks mean smaller businesses have no choice other than to connect to them as subservient partners.

Analysts at stockbroker Truist predicted Muse could add $US28.5 billion to Meta’s company revenue by 2030.

Other analysts are sceptical, with Oppenheimer suggesting such success would require Meta to acquire around 115 million subscribers paying for the product at $US20 a month.

The cost of providing the energy-intensive computing power required by AI agents, means many analysts think most of the major players will eventually need to charge users or businesses to make the technology profitable.