The third era of the web is here

Hopes were initially high that the world wide web would be a decentralizing and disintermediating technology. 

“One trend that I see is that it’s the individual who replaces the big organization in many of our lives,” civil libertarian and Grateful Dead lyricist John Perry Barlow wrote in 1996.

In that first era of the web, the big organizations most likely to be replaced were publishers.

“Writers and readers can now connect to one another directly,” internet evangelist Richard Seltzer wrote in 1995. “Anyone can be a publisher.”

Soon, seemingly everyone was: Blogs, podcasts, and newsletters proliferated, fulfilling the web’s early promise of breaking down corporate barriers.

“It is by now evident to everyone that the days of the press, especially major newspapers, playing the role of gatekeeper are gone,” The Washington Post lamented in 1998.

The Post turned out to be one of the few survivors. Most publishing concerns discovered that their gatekeeper status was based not on the content they created, but the printing presses and delivery trucks they used to distribute it.

When the web made distribution free, content became abundant. You no longer had to wait for the newspaper to be delivered; whatever information you wanted was available online, whenever you wanted it.

This created a new problem: discovery. Amid the explosion of content, people needed a way to find what they wanted.

At first, portals like “Jerry and David’s Guide to the World Wide Web” manually curated links to websites for us. As the links proliferated, content became overabundant, making attention the scarce resource. The companies that could aggregate the most users — and match them with the limitless supply of content — became the gatekeepers of the internet: Facebook, Google, Netflix.

Before the internet, the hardest problem was getting content to users: It required delivery trucks, TV stations, and movie theaters. After the internet, the hardest problem was getting the users.

The power of the internet’s new gatekeepers, Ben Thompson explains, comes from controlling the users. Once an aggregator amasses sufficient demand, content suppliers have to come to them on their terms “even though they hate it.”

This is Thompson’s Aggregation Theory, which he used to explain the power dynamics of the second era of the world wide web.

Now, however, we’re entering a third era — an agentic one.

The second era began when the web made viewing content on the internet abundant. The third era began when Muse.ai made doing things on the internet abundant.

The web made information abundant, but getting things done remained laborious. To book a flight, order groceries, or file a form, you still had to find the right website or app for the task, log on, and do a lot of clicking and typing.

As of September 8, however, a Muse agent can do that for you. Just give it your login details and it can navigate the web on your behalf.

Here’s a live shot of mine filing an insurance claim that I wouldn’t have otherwise have bothered with:

I got $198 back! Amazing.

And instructive.

“This is a replay of what happened with websites,” Thompson says. “What is happening with agents is that the ability to do stuff is becoming abundant.”

Websites made content abundant, which created a new set of gatekeepers — the tech giants that aggregated our attention by solving the problem of discovery.

The new problem to solve, Thompson says, is “inspiration:” figuring out what to have our agents do when they can do almost anything.

“The companies who solve inspiration will gain power over every entity that has things that need to be done.”

One thing we might be inspired to do is start a bank run.

“Muse and similar agentic AI assistants could soon sweep household cash automatically into accounts paying 3.3% to 5.0%, instead of the 0.1% national average on checking accounts,” economist Torsten Slok predicts.

“If every household used AI agents to optimize the return on their cash balances,” Slok adds, “banks could lose a large share of the cheap deposits they rely on to make loans, which would be a problem for the entire financial system.”

Slok hinted this could cause the ultimate bank problem with the title of his short note: “Is an agentic bank run coming?”

The answer is almost certainly not.

I, for example, will not be asking Muse to sweep the checking account at my bank into a fintech for the sake of 3 or 4% of additional yield. That money is working capital, not investment capital and I don’t need to optimize it.

And even if I did, I wouldn’t ask a fallible, possibly misaligned AI agent to do it for me. However cute and cuddly the avatar, I’ll continue to use my banking apps myself, thank you very much.

But there are plenty of apps and websites I will give an agent access to. Like my fantasy baseball one, for example. If my second baseman gets hurt next season, I won’t be opening the CBS Sports app to search for a replacement. I’ll ask Muse to log on to the app and do it for me.

Along with a lot of other things. Just as Google and Facebook became the interface to the web’s abundant content, agents will become the interface to the web’s abundant capabilities.

This, Thompson concludes, is the new — and ultimate — prize in technology: becoming the interface between us and everything we want to do online.

— Byron Gilliam

_{Brought to you by:}

Meridian 2026 is Stellar's annual gathering for the institutions, fintechs, and developers putting financial infrastructure onchain.

Join them October 28-29 at Convento do Beato in Lisbon for two days on tokenization, payments, and what it takes to run them in production.

Register today with Blockworks10 for 10% off.