--- title: "The Impact of AI Agents on the Digital Economy (2026)" description: "34 hypotheses on what AI agents do to aggregators, marketplaces, pricing, brands, banks, insurance and jobs once buyers stop being busy or forgetful." url: "https://www.maltelandwehr.de/articles/ai-agents-digital-economy/" lang: en --- # The Impact of AI Agents on the Digital Economy: 34 Hypotheses By Malte Landwehr, published 28 October 2026, last updated 28 October 2026 ## Summary AI search moved the buying decision from the website into the chat. Agents **take buying decisions out of human hands**. That hurts every business model that depends on humans being busy, forgetful, or not well informed. Pure routers like **price comparison sites and metasearch are hit hardest**. Marketplaces that handle payment, delivery, and trust have a chance if they **evolve into service companies**. Many products and **subscriptions will get more expensive**, because the people who never used them stop paying for them. And many jobs that exist to sell, negotiate, or answer the phone will move to agent-to-agent communication. Contents 1. [Part 1: Shopping, Marketplaces, and Aggregators](#shopping-marketplaces-aggregators) 2. [Part 2: Pricing and Business Models](#pricing-business-models) 3. [Part 3: Brands, Products, and Reviews](#brands-products-reviews) 4. [Part 4: Banks and Insurance](#banks-insurance) 5. [Part 5: Consumers and Society](#consumers-society) 6. [Part 6: Jobs](#jobs) 7. [Bottom Line](#bottom-line) 8. [Sources](#sources) In September 2026 Meta launched Muse, an AI agent that can search, compare, and book my travel directly with airlines and hotels. The stock prices of Booking, Expedia, and Airbnb immediately dropped by 6%.[Source \[1\]](#source-1) The reason is easy to see. Imagine turning ten hours of comparing hotels, prices, and cancellation policies into one prompt: “*I want to spend two weekends in Paris with my wife. Direct flights after 10am. Quiet apartment near the center. Budget 5,000 Euro.*” The agent does the research. You never visit Booking, Expedia, or a price comparison site. I believe agents will fundamentally change how aggregators, price comparison services, and marketplaces work. Beyond that, a lot of business models only work because humans are busy, forgetful, or not well informed. Agents are none of that. Here are my 34 hypotheses on what happens when agents start fulfilling tasks for us, grouped by who is affected. Some of them I am confident about. Others are educated guesses. ## Part 1: Shopping, Marketplaces, and Aggregators ### 1. The Buying Decision Moves From the Website to the Agent Before AI, humans decided on Booking, Skyscanner, or idealo which offer they wanted and clicked through. User, merchant, and aggregator all profited. With AI search, the decision about what to buy and where is made in the chat. Visitors land on a product page and only need the merchant to finish the transaction. With agents, the shopper does not land on a website at all. The agent gathers information on many websites and buys where it finds the best conditions. ### 2. Product & Price Comparison Sites Become Obsolete Comparison sites save humans time and give an overview of scattered inventory. That service attracts humans. Merchants pay a small fee in exchange for access to these human shoppers. Agents will massively disrupt this business model and make the current aggregators obsolete. An agent is much less reliant on a middleman to save time. If an agent uses an aggregator, it will always double check the hotel or retailer directly. There is no incentive to click on the affiliate links provided by a price comparison website. For very crowded markets - like hotels or online shopping - aggregators still offer something valuable for agents. Because an agent cannot check every hotel or every online shop. But for markets like flights or rental cars, where it is feasible to check every single vendor individually, the value of the aggregator trends towards zero. In these crowded markets, I see two possible paths for aggregators: 1. Consolidate demand by offering the best possible agent experience and charge merchants a tiny fee. The problem is that you can never have the best inventory with this route and premium merchants have no incentive to participate. 2. Consolidate supply by listing all merchants for free - even against their will. Then charge the agents a small usage fee. Neither seems like a sustainable business model on its own. ### 3. Marketplaces Survive Only as Service Companies Marketplaces that handle the purchase are in a better position than pure routers. Someone still has to handle payment, shipping, returns, and refunds. Still, many marketplaces will disappear. And the survivors will shift from tech to service. A travel marketplace that refunds you when the apartment doesn’t match the photos has something an agent cannot replace. ### 4. Ads and Paid Placement Stop Working Marketplaces and comparison sites make their money by selling placements and retail media ads because humans click one of the first results. An agent doesn’t care about a product that is artificially ranking higher. It reads the whole list. Example: 64% of the products Amazon’s Rufus recommends do not rank in Amazon’s top 10 organic results.[Source \[2\]](#source-2) ### 5. Companies That Block Agents Lose Some companies will fight agents and lose massively. Conglomerates with their own agents can block other agents for a while: Amazon with Rufus, Meta with Muse, Google with its own agent. A car rental company, cinema, or hotel website that blocks agents is doomed. ### 6. Experience Shopping Stays Human Some products we - partially - buy for the experience, like a Rolex or Ferrari. Here, people are unlikely to use agents. ## Part 2: Pricing and Business Models ### 7. Every Price Match Guarantee Gets Claimed Humans rarely claim a price match guarantee. An agent claims it every time. It picks the provider with the best service, makes them match the cheapest price, and keeps checking after the booking. That turns the guarantee into a guaranteed discount. At some point, Booking and others will have to stop offering price matching. ### 8. Every Public Coupon Becomes a Permanent Discount Agents hunt for every bargain. A brand that sometimes discounts will only sell during the discount. Coupons are price discrimination that only works because few people use them. Agents apply and combine every coupon they can find. So a public coupon is a permanent discount. Companies that rely on coupons will have to rethink their business model. ### 9. Bait-and-Switch Incentives Stop Working Banks, credit card companies, and electricity providers lure new customers with signup bonuses and big first-year discounts. Sometimes offering hundreds of Dollars in cash! This only works because most people forget to switch again. Agents will chase every signup bonus and switch every year. Bait-and-switch incentives stop paying off and probably disappear completely. Unless a brand is sure the new customers will stick because of superior service. ### 10. Sales Events Boom, Then Shrink Agents learn when the price is lowest and wait. That first causes a boom in Black Friday-style sales. Over time, companies realize agents never buy at full price after a big drop, and they stop deep discounts. ### 11. Inactive Members Stop Subsidizing Active Ones Gym members who never show up pay for those who train five times a week. Agents will cancel unused subscriptions: streaming, software, cloud storage, phone plans, learning apps. Then active users pay the real cost. As a consequence, gyms, Netflix & Co get more expensive. Usage-based pricing is the alternative. And there power users pay much more. ### 12. Add-Ons With Negative Expected Value Stop Selling The insurance upsell at the car rental counter has a negative expected value for the consumer. Uninformed consumers buy it - and boost the car rental company’s margin. An agent will never buy it. When nobody buys the add-on, the base price has to go up. ### 13. Free Cancellation Disappears Free cancellation will disappear. Otherwise my agent could reserve five restaurants every evening and cancel four. Cancellations will either cost a fee or be tied to a reputation based on a verifiable, possibly pseudonymous, identity. Cancel too often, and you cannot participate anymore. ### 14. Generous Service Policies Disappear Generous return and refund policies only work because few people max them out. Agents will claim everything a policy allows. Those policies will disappear. ### 15. Long-Term Commitments Become the Strongest Lock-In If an agent knows the user pays for Prime or Costco, it will use that membership for every small purchase. So the battle is still for the irrational human. Companies need to convince humans to make long-term commitments, like a 36-month membership with no option to cancel. Once the human has signed, the agent will make use of it. ### 16. Companies Rush to Lock In Long-Term Deals Before agents spread, companies will rush to lock in multi-year contracts and lifetime deals. The window is now, while humans still make these decisions themselves. ### 17. More Custom Bundles Bundles are already hard to compare. With agents, we will see many more of them: custom bundles and custom versions of products that exist nowhere else, so there is nothing to compare them to. ### 18. Every Bundle Includes an Item for Agents Bundles will include at least one item that is specifically targeted at convincing agents, not humans. ## Part 3: Brands, Products, and Reviews ### 19. Perceived Quality Beats Brand Advertising Agents reward perceived product quality, which they infer from reviews, specs, and return rates. TV ads from their childhood don’t influence them. While agents are not 100% rational,[Source \[3\]](#source-3) they are a lot more rational and a hundred times better informed than the average human shopper. ### 20. Fewer Brands, or More Long Tail Here I see two scenarios: 1. **Concentration.** If agents do most of the buying, they might all pick the same product. Only that product gets new reviews, creating a loop. So we end up with fewer brands. 2. **Long tail.** Agents find niche products that fit each user perfectly, and the long tail grows to sustain more niche brands. I’m honestly not sure which scenario wins. In both cases undifferentiated brands living off brand equity built in the past, will disappear. Most European car brands are a good example for this. ### 21. Reviews Get Incentivized If I didn’t buy a product myself, why would I review it? Reviews are an unsolved problem in the age of AI agents. One solution is incentives: rewards for agents that write reliable reviews, free access only for people/agents who write reviews, or paid human reviewers. Verified review providers like Cimenio will massively profit. ### 22. Product Pages Are Written for Agents Product pages will be written for agents too: exact specs, prices, delivery times, and return conditions. In a format and structure agents can understand. ### 23. Brands Fight Over the User Context Brands will compete to influence what agents know about their users. Email newsletters already influence Google AI Mode.[Source \[4\]](#source-4) WhatsApp, text messages, Messenger, Instagram DMs, and Slack come next. ## Part 4: Banks and Insurance ### 24. Money Never Stops Moving Agents will move money to wherever the interest is highest, so bank runs become the norm. Banks will answer with lock-ins and withdrawal limits. Money that is available anytime becomes a premium product. ### 25. Insurance Gets Rethought Insurers profit from people being over-insured. Agents will cancel unneeded policies and flock to any policy that is a good deal for their user. Insurers will price per micro-segment, and the business has to be rethought. ## Part 5: Consumers and Society ### 26. Lazy Consumers Win, Smart Consumers Lose Agents are good for lazy consumers and bad for smart ones. Today, smart shoppers profit at the expense of lazy ones. With agents, everyone shops smart. The smart shopper’s edge is gone. This hits people the hardest who are poor and smart. It also makes a permanent underclass more likely. ### 27. Your Agent Becomes a New Class Divide Rich and tech-savvy people will have agents that work for them. Everyone else gets free, ad-funded agents built for suppliers and advertisers. If you are not paying, you are the product. This could make wealth disparity more extreme, cementing the permanent underclass. ### 28. Courts Decide Who Is Liable Who is liable when an agent makes a terrible mistake? The user or the creator of the agent/harness? Expect court cases and laws. ## Part 6: Jobs ### 29. Call Centers Become Agent-to-Agent Call centers can’t handle the volume of agent calls, so they go agent-to-agent. Talking to a human becomes a premium service. ### 30. Insurance Sales Reps Lose Insurance sales reps lose. Especially in over-insured markets like Germany. ### 31. Haggling Becomes Agent-to-Agent Agent-to-agent negotiation replaces most haggling jobs. From car dealers to real estate agents. ### 32. Procurement Goes First B2B buying goes agentic first. When only price and reliability matter, buyer and merchant agents negotiate directly. Purchasing departments for standard goods shrink first. ### 33. Generic Sales Staff Disappear Generic sales staff in most stores disappear. Sales staff who create an experience for Rolex or Ferrari buyers gain value. ### 34. New Jobs Appear Agent experience (AX), the agent version of UX, and agent relations, the PR for agents, become real jobs. ## Bottom Line AI search moved the buying decision from the website into the chat. Agents take buying decisions out of human hands. That hurts every business model that depends on humans being busy, forgetful, or not well informed. Pure routers like price comparison sites and metasearch are hit hardest. Marketplaces that handle payment, delivery, and trust have a chance - if they evolve into service companies. Many products and subscriptions will get more expensive, because the people who never used them stop paying for them. And many jobs that exist to sell, negotiate, or answer the phone will move to agent-to-agent communication. ## Sources 1. 24/7 Wall St.: [Travel Booking Stocks Tumble as Muse Threatens to Bypass Them](https://247wallst.com/investing/2026/09/23/travel-booking-stocks-tumble-as-muse-threatens-to-bypass-them-expedia-falls-7-airbnb-drops-6-booking-holdings-sinks-5/), September 2026. [↑](#ref-1) 2. Autopilot & The Order: [The Third Shelf](https://www.autopilotbrand.com/thirdshelf), January 2026. [↑](#ref-2) 3. Jan Ehrlinspiel, Peec AI: [AI assistants judge your company by its name](https://peec.ai/blog/ai-assistants-judge-your-company-by-its-name-i-measured-how-much), August 2026. [↑](#ref-3) 4. Garrett Sussman, iPullRank: [Your Inbox Might Be the Next AI Search Signal](https://ipullrank.com/google-personal-intelligence-experiment), May 2026. [↑](#ref-4) ![Malte Landwehr](https://www.maltelandwehr.de/assets/img/malte-landwehr-author-192.jpg) [Malte Landwehr](https://www.maltelandwehr.de/) is an expert in Enterprise SEO and AI search with 20+ years of experience. [Speaker profile](https://www.maltelandwehr.de/speaker/)