Editor’s note: Today is the last day to save $500 on a RiskHedge Venture membership. When you join, you’ll also get Stephen’s Crypto Masterclass for free. Go here to learn more before the offer expires.

The artificial intelligence (AI) trade is moving into its next phase.

For the past few years, the biggest winners in AI were the companies supplying the picks and shovels.

GPUs. Memory. Networking equipment. Data center power and cooling. And everything else needed to build the enormous infrastructure behind artificial intelligence.

We got in early on this boom in Disruption Investor, locking in triple-digit gains on names like Nvidia (NVDA), Micron Technology (MU), and Tawain Semiconductor (TSM) along the way.

That infrastructure trade is still alive and kicking. AI companies continue to spend huge sums building out capacity. And the world is likely to consume far more computing power in the years ahead.

But every new technology eventually moves from the infrastructure phase into the application phase.

We’re starting to see that with AI.

The apps built on top of all that infrastructure are beginning to influence which companies win and lose in the stock market.

• On September 8, Meta Platforms (META) launched Muse.

Muse quickly shot to No. 1 on the App Store. It also sent Meta’s stock surging 27%.

Muse is essentially AI for everyday life.

For the past few years, we’ve mostly used AI as something we talk to. Muse points to what comes next: AI agents that can actually do things for us.

A chatbot tells you how to do something. An agent goes and does it for you.

Say you’re planning a weekend trip. A chatbot can recommend hotels, compare neighborhoods, and tell you where to eat. 

An agent can search the flights. Find a hotel within your budget. Make the reservations. Add everything to your calendar. And send you the itinerary when it’s done.

That seemingly small jump from giving answers to acting will have enormous consequences.

I’ve been using agents myself. Last week, I got access to Instinct. It’s a personal AI agent that I can message through iMessage or WhatsApp. I can even send it a voice note and tell it what I need done. 

One of my first requests was to track down and claim missing airline miles. It went off and did the legwork for me (more on Instinct Friday).

• Agents will disrupt entire industries.

Take online travel. Hotels often pay hefty commissions or offer discounted inventory to platforms like Booking Holdings (BKNG) and Expedia Group (EXPE). They help put rooms in front of travelers. 

But an AI agent could bypass much of that process. It can search hotels directly. Compare prices across the web. Weigh reviews on Google. And book the best option in seconds. The need for middlemen disappears.

Then there’s software.

A huge part of the software industry is built around charging companies for every employee who uses a product. Buy 500 seats. Pay for 500 users.

Agents will turn that model on its head, too. 

If one agent can do work that once required several employees across multiple software tools, companies may start asking why they’re still paying for hundreds of seats. 500 seats could shrink to one agent.

Online advertisers are also at risk.

They make money by putting ads in front of humans as they search, browse, shop, and compare options. An AI agent can skip much of that process.

Say you want to buy a new coffee machine. Instead of searching Google... clicking through several websites... reading reviews... looking through ads... and comparing prices yourself... you could simply tell your agent what you want.

The agent can do the digging for you and come back with the best option. And if the agent is doing the searching, there’s no one left to see the ads.

That changes the economics for companies that rely on human attention as the toll booth, like Google (GOOGL) and Amazon (AMZN).

• Agents will create winners, too.

An agent that checks prices across 30 websites, schedules meetings, monitors your inbox, searches for flights, and manages routine work burns through far more computing power than someone typing a handful of questions into ChatGPT.

That creates another tailwind for many of the same AI infrastructure winners we’ve been investing in so far.

But the opportunity goes well beyond infrastructure.

Think about all the businesses that will benefit once agents start doing more of the work for us. Payment companies that help agents transact. Cybersecurity firms that verify what they’re allowed to do. Software platforms that become the tools agents use to get things done. 

Even smaller brands could benefit if agents make it easier for customers to discover the best product instead of simply the biggest name.

That’s why we’re building an “Agent Index.” 

It’ll contain two baskets. One of companies and business models positioned to profit from the agent era. And another made up of those most exposed to disruption.

We’re digging into both sides now. We’ll be sharing that research with you in the weeks ahead. Stay tuned.

Stephen McBride

Chief Analyst, RiskHedge

PS: A quick heads up before you go: Our RiskHedge Venture sale ends tonight. If you’ve been thinking of joining me, there’s still time to act before this deal is taken offline. I made Venture to help readers understand how I approach crypto, what I look for in great crypto businesses, and how to invest in this market without getting sucked into all the hype. Check out the details here.