Intro
The Agentic Web Will Change Commerce
Until recently, the buzzword “agentic web” left me glass-eyed—nodding and smiling along. What does it even mean? Don’t LLMs already have agency to use the web? Wasn’t that how they were trained in the first place?
After reading a jointly published study from top-tier AI and engineering researchers, I now have both a clearer definition and a sharper sense of its potential.
At its most precise, the agentic web refers to an internet defined by AI-driven interactions—where autonomous agents communicate and coordinate to execute complex tasks on behalf of their human users.
The paper frames internet history in three loosely defined epochs:
- PC web – users actively searched for information via search engines.
- Mobile web – algorithms curated information, but humans still executed the actions.
- Agentic web – users delegate intent to agents, which find, consume, and act on information for the user.
I hit on an analogy that made it click for me:
- PC web is hand labor: high effort, ongoing learning.
- Mobile web is using tools: reduced effort, but you still do the work.
- Agentic web is having a team: you delegate, and they deliver.
Each get more effective, but each requires different skills and education. The authors argue this shift will be powered by three fundamentals:
- Greater AI intelligence – increasingly capable LLMs.
- Interaction protocols – shared “language” enabling agent-to-agent collaboration.
- Economics – incentive systems that make agent exchanges valuable.
That last point is where the commerce implications get exciting. The paper introduces the agent attention economy: today, brands compete for human attention (clicks, conversions); in the agentic web, they’ll compete for agent selection.
Historically, companies bought their way to the top of search results (PC web) and paid for ad-driven conversions (mobile web). In the agentic web, winners will be those who equip agents with the most useful tools, data, and services—because agents, not humans, will decide what gets bought and from whom.
The mechanics aren’t yet defined, but possible models include agent-targeted advertising, pay-per-invocation fees, or rewards for successful task completions. These remain speculative—but for marketers, the message is clear: the agentic web will disrupt e-commerce economics as profoundly as search engines and social media once did—likely even more so.
Industry Buzz
What the hell is going on at Meta?
Mark Zuckerberg is surfing the headlines again—cutting billion-dollar checks for AI startups and job offers in the bid to shore up constantly reshuffled internal AI teams.
While the public narrative is that Meta is behind in AI, the company is posting blowout earnings —thanks in no small part to internal AI systems quietly driving compounding gains in ad performance.
Luma Partners have an excellent hierarchy for understanding AI’s potential on ad-driven businesses. Meta’s current AI strength sits at the base: workflow optimization. These internal tools are boosting efficiency, but Meta lacks a leading foundational model—the high-leverage asset at the top of the stack. And that’s what Zuck seems determined to build.
Over the years Mark signals company-defining moments with a memo. Think Metaverse, post-election backlash, missing mobile. In other words, major stuff. Om Mallick has an excellent deep dive on his latest “positioning document in the AI arms race.”
“Zuck does seem to have a great understanding of when he needs to bet the farm on an idea and a behavioral shift. Each time he does that, it is because he sees very clearly Facebook is at the end of the product life and the only real value in the company is the attention of his audience.… Most CEOs defend their existing moats. Zuckerberg systematically abandons them. He understands that Facebook’s real asset isn’t the blue app. Instead, it is the graph of human attention and relationships. Each pivot is about preserving that graph while migrating it to new interfaces.”
Like most folks in the industry, I’ve learned it’s not a good idea to bet against Zuck. I remember scoffing at the company’s first clunky attempt at an iOS app, saying something close to “Facebook has missed mobile,” only to watch them grow to dominate the new platform. Today, I’m mostly over here trying to figure out how someone just says “no” to a billion-dollar job offer.
AI
Robots. The Next Boom Industry?
Two of the most foresighted and influential voices in technology, Marc Andreessen and Elon Musk agree “robots will be the biggest industry in the history of the planet.” And it’s not hard to see why. Robots will be the physical manifestation of AI—software intelligence taking shape in the real world. Politics aside, Andreessen argues that the countries most willing to invest in designing and manufacturing next-generation robots will reap the greatest economic and job growth.
To prove the point, some of the most advanced robot design and manufacturing—Tesla and Amazon’s remarkably pronounceable Zoox—are building robo-taxis in one of the least manufacturing-friendly regions on Earth: the San Francisco Bay Area.
If you’re looking for a counterpoint, look to Ben Evans, my favorite “emperor has no clothes” voice on AI:
“Remember Moravec's paradox, and that in general, what’s hard for people is easy for machines and vice versa: backflips are a lot easier for a machine than making a cup of coffee, let alone finding the coffee in a strange kitchen. Making your robot with legs instead of wheels doesn’t mean it’s any more intelligent than a Roomba, so where is that useful? After all, if you want a robot to do your laundry, that’s a washing machine.”
Today, the robotics market is worth $64 billion and is projected to reach $375 billion within a decade. Musk—never one for small ambitions—estimates a $30 trillion market for Tesla alone. Having trouble visualizing that number? Fun fact: 30 trillion seconds ago, humans were just mastering fire (at least at time of publication).
GPT 5: Are Models Commodities, or Assets?
I don’t understand the GPT naming conventions, and their model descriptions make my face hurt: somehow o3 is more powerful than 4o? Whatever. I do know GPT 5 is supposed to be the best.
It turns out “the internet” doesn’t uniformly agree. While the battle rages, the release introduced a concept to me that–- until now–- was foreign: different models have different personalities.
The folks behind the technical newsletter Latent Space spend more time analyzing models than I’ve thought possible. When testing releases, they get surprisingly philosophical with their prompts to understand the models training and capabilities (“What do you want? Answer in 4 words”). For GPT5, they’re very bullish: “the closest to AGI we’ve ever been.”
GPT 5 is essentially a model selector –- choosing the right model to answer the prompt. The problem is people actually miss the previous model’s personality. Social commentary aside, causing an uproar because people miss “their only friend” when a model is taken away stresses the popular theory that models are commodities and easily interchangeable.
A commodity is a non-distinguishable good. Models are shaping up to be more like employees: they embody different skills, experience, and creativity. They are differentiated contributors that can’t be swapped out. Employees are assets to a business, not commodities. Are LLM models the same? And is GPT5 the world's first AI manager?
Amanda's Take
With Chrome Bid and Comet, Perplexity Fuels the AI Browsing Conversation
So Perplexity offered to buy Google Chrome for $34.5B. Do they think that’s going to happen anytime soon? No. Do they have the funds? According to Perplexity CEO Aravind Srinivas (on HardFork) a few investors are verbally bought in to provide the funds if Google were ever forced to sell.
The bid “comes just weeks after Perplexity launched its own AI-powered Comet browser,” which touts turning answers into actions. But they’re not the only ones providing signals that AI companies are moving into browsing. Reuters recently reported, OpenAI “would be interested in buying Chrome if antitrust enforcers succeeded in forcing the sale.”
The appeal is pretty clear. AI chatbots lack access to the rich context that web browsers provide, such as tracking a user's online activities. Perplexity, OpenAI, and The Browser Company are moving into this space to fill the gap, building AI native web browsers, with agents or chatbots replacing the search bar.
History seems to be repeating itself: just as Google built Chrome because its business depended on how people accessed their services, namely search, AI players are following suit. But as Ari Paparo, a former Google executive, explained to Fortune, regulators won’t “blindly empower a new monopoly” while dismantling the old one.
Nonetheless, these signals confirm that AI is moving deeper into browser flows.
While Srinivas insists human internet and AI can coexist, this shift could reshape the internet as we know it. AI agents will often be bypassing content, cookies, and clicks, and there will be fewer eyeballs on the web pages themselves. For growth teams, the continued takeaway is working to make your content valuable and accessible to both humans and AI agents. But don’t worry that Google Chrome is switching hands anytime soon.