A boutique hotel in Banff pays Booking.com between 15% and 25% of every reservation that comes through the platform. On a $350-per-night room booked for three nights, that's somewhere between $157 and $262 going to Booking.com for the privilege of being found. The hotel doesn't love this arrangement. But the alternative — being invisible to the millions of travellers who start their search on an OTA — is worse.
A restaurant in Calgary's Kensington neighbourhood pays Google roughly $3-5 per click on its search ads. Not per customer — per click. Most clicks don't convert. The restaurant might pay $400 in a month for 100 clicks that produce 15 actual diners. That's about $27 per customer in acquisition cost, before the customer has ordered a single appetizer.
A dog daycare in Vancouver pays Yelp $400 per month for enhanced placement — a featured listing, better positioning in search results, the ability to add a call-to-action button. Whether the daycare gets two new clients from Yelp that month or twenty, the fee is the same. The daycare is paying for visibility, not results.
These are the economics of platform-mediated discovery. The platforms control the funnel — the place where consumers go to find businesses — and they extract rent from every business that wants to be visible in that funnel. The rent takes different forms (commission, cost-per-click, subscription), but the underlying dynamic is the same: you pay to be found, and the platform captures a significant share of the value you create.
How agent-mediated discovery changes the math
AI agents don't work like platforms. They don't curate a marketplace that businesses pay to be listed in. They don't rank businesses based on ad spend. They don't insert themselves between the customer and the business and extract a percentage of every transaction.
An AI agent works for the consumer. It queries data sources, evaluates options against the consumer's specific needs, and recommends the best match. The agent's incentive is to make the best recommendation — because the consumer judges the agent on the quality of its suggestions, not on which business paid the most to appear first.
This changes the discovery model fundamentally. When a business provides structured data to a data layer that agents can query, and an agent recommends that business to a consumer, there is no platform in the middle collecting 20% of the transaction. The data layer charges a fee when a match is made — a per-transaction cost that is fundamentally different from the subscription fees, ad spend, or commissions that define platform economics.
The difference matters because it changes who captures the value. In platform economics, the platform captures a growing share of the value as it becomes more dominant. In agent-mediated economics, the value stays with the business because the agent is working for the consumer, not monetizing the business.
The math for a hotel
Let's walk through what this looks like for a 30-room boutique property in Canmore, Alberta.
Current platform costs: The hotel books roughly 40% of its rooms through OTAs (Booking.com, Expedia). Average nightly rate: $280. Average stay: 2.3 nights. That's about $644 per booking. At a 20% commission, the hotel pays roughly $129 per OTA booking. Over a year, with roughly 2,500 room-nights booked through OTAs, the hotel pays approximately $140,000 in OTA commissions.
The hotel also spends about $2,000 per month on Google Ads to drive direct bookings, and $500 per month on TripAdvisor placement. That's another $30,000 per year in advertising costs.
Total annual cost of platform-mediated discovery: roughly $170,000.
Agent-mediated alternative: The hotel structures its data — real pet policy, current availability, differentiators like trail access and mountain views from specific rooms, seller intent for midweek gaps. This data lives in a queryable layer that AI agents access. When an agent matches a traveller to this hotel and the traveller books, the data layer charges a transaction fee.
Even if agent-mediated bookings replaced just half of the hotel's OTA volume, the savings would be substantial. The transaction fee on a data layer match is a fraction of a 20% OTA commission. If the hotel shifts 1,250 room-nights from OTA-mediated to agent-mediated discovery, at a dramatically lower per-transaction cost, the annual savings could reach $80,000 to $100,000 — money that goes back into the property, the guest experience, or the owner's pocket.
The math for a restaurant
A mid-range restaurant in Toronto doing 150 covers a night spends roughly $3,000 per month on Google Ads, $600 per month on Yelp advertising, and another $1,000 per month on Instagram promotion and food photography for social media visibility. That's $55,200 per year in discovery costs.
The restaurant's average cover is $55. Its food cost is 30%, labour is 30%, and overhead is 25%. That leaves a 15% margin — about $8.25 per cover. When the restaurant pays $4 per Google click and converts one in five clicks, it's paying $20 to acquire a customer worth $8.25 in margin on the first visit. The math only works if the customer comes back — and most restaurant customers acquired through ads don't.
Agent-mediated discovery resets this equation. The restaurant structures its data: current menu with pricing, noise level, private dining capacity, dietary accommodations (the chef trained in Paris, does exceptional gluten-free French cuisine — that's a differentiator), real-time table availability, and a seller intent signal that they want more Thursday and Sunday night reservations. An agent matches a diner to this restaurant not because the restaurant paid more, but because the data says it's the right fit.
The per-match transaction fee is a fraction of the $20 per customer the restaurant currently pays through Google Ads. And the match quality is higher — the agent sent a diner who specifically wanted what this restaurant offers, not someone who clicked an ad while casually browsing "restaurants near me."
The math for a service business
A physiotherapy clinic in Edmonton with four practitioners spends $1,800 per month on Google Ads targeting keywords like "physiotherapy Edmonton" and "sports injury rehab Edmonton." It pays $500 per month for an enhanced listing on a healthcare directory. It budgets $300 per month for social media management. Total: $31,200 per year in discovery costs.
The clinic's average new patient generates $1,200 in revenue over a treatment course (ten sessions at $120). Its Google Ads convert at about 3% — for every 100 clicks at $4 each, the clinic gets three new patients. That's $400 in ad spend per $1,200 patient, or a 33% acquisition cost. Better than the hotel's OTA commission, but still a significant drag on margin.
With agent-mediated discovery: The clinic structures its data — each practitioner's specialization ("post-ACL reconstruction, return-to-sport protocols for competitive athletes"), current availability by practitioner, accepted insurance plans, and seller intent ("accepting new patients for shoulder rehabilitation, especially post-surgical"). When an agent matches a patient recovering from ACL surgery to the specific practitioner who specializes in exactly that, the match quality is incomparably better than a Google Ad click — and the cost is a fraction.
The patient gets a practitioner who actually specializes in their condition, rather than the first clinic that appeared in a Google search. The clinic gets a patient who matches their expertise and is likely to complete the full treatment course. And neither party paid 33% of the transaction value for the privilege of being connected.
Commission-free doesn't mean free
Let's be precise about the terminology. "Commission-free" doesn't mean there's no cost to agent-mediated discovery. It means the cost structure is fundamentally different.
Platform commissions are a percentage of revenue — the more successful the transaction, the more the platform takes. A hotel that raises its rates pays a higher absolute commission. A restaurant that sells more expensive wine pays more to the platform that sent the customer. The platform's incentive is to take a growing share of the business's success.
Agent-mediated discovery costs are per-transaction fees tied to the match, not the revenue. The data layer charges when a match is made — when an agent queries the data, finds a business that fits the consumer's needs, and facilitates a connection. The fee is for the data and the match, not a percentage of what happens after.
This distinction matters enormously. A per-transaction fee means the business's incentive to increase order value, build loyalty, and maximize lifetime customer value is undiminished. When the hotel upsells the suite, when the restaurant sells the expensive wine pairing, when the physio clinic retains the patient for a full treatment course — the business keeps that value. No platform is extracting 20% of the upside.
Commission-free means aligned incentives. You pay for the match, not for the outcome. The better you do after the match — higher ticket, more repeat visits, stronger lifetime value — the more value you keep. That's the opposite of how platforms work, where your success is the platform's revenue opportunity.
The reinvestment opportunity
The most important consequence of commission-free economics isn't the cost savings themselves — it's what businesses do with the savings.
The Canmore hotel saving $80,000 per year in OTA commissions can renovate rooms, add amenities, hire better staff, or reduce rates to attract more direct bookings. Each of these investments improves the guest experience, which generates better reviews, which makes agents even more confident in recommending the property. The savings become a flywheel.
The Toronto restaurant saving $30,000 per year in advertising costs can invest in better ingredients, bring in a pastry chef, or extend hours on slow nights with programming that agents can recommend. The Edmonton physio clinic can add a new practitioner, invest in specialized equipment, or lower session rates for patients paying out of pocket.
When platforms extract less, businesses can invest more in what actually matters: the quality of the product or service. And in an agent-mediated world, quality is what determines recommendation — not ad spend, not review manipulation, not SEO tricks. The businesses that reinvest their platform savings into genuine quality improvements will see those improvements reflected in better agent recommendations, creating a virtuous cycle that platform economics never allowed.
The era of paying a quarter of your revenue for the privilege of being found is ending. Agent-mediated discovery lets businesses compete on what they're actually good at — not on how much they can afford to pay a platform gatekeeper. That's not just better economics. It's a better system.