In 1995, the Yellow Pages was the most profitable business in local advertising. Every plumber, dentist, and restaurant paid for a listing because that's where customers looked. The Yellow Pages company didn't provide the services. It didn't even evaluate them. It just sat in the middle and collected rent on attention.
Google did the same thing, better. Instead of alphabetical listings, it offered ranked results. Instead of paying for ad size, businesses paid per click. The middleman changed, but the structure didn't — a platform that sits between consumers and local businesses, extracting value from the connection.
Now the middleman is changing again. AI agents are becoming the new intermediary layer between consumers and local commerce. And this time, the shift is more fundamental than swapping one directory for another.
The middleman pattern
Every era of local commerce has been defined by its middleman:
1960s-1990s: Phone directories. Consumers looked up businesses in the Yellow Pages. Businesses paid for display ads. The bigger your ad, the more calls you got. The Yellow Pages company printed books and collected advertising revenue. Peak revenue: $14 billion in the US alone.
2000s-2010s: Search engines. Consumers typed queries into Google. Businesses optimized their websites and paid for ads. Google ranked results by relevance (and willingness to pay). Peak: Google's ad revenue hit $224 billion in 2024, a substantial portion from local and shopping queries.
2010s-2020s: Platforms and marketplaces. Yelp for restaurants. Zillow for real estate. AutoTrader for cars. Booking.com for hotels. Each platform owned a vertical and charged businesses for leads, listings, or transactions. Businesses had to be on the right platforms to be found.
2025+: AI agents. Consumers tell an AI agent what they need. The agent queries available data sources, reasons over the options, and returns a recommendation — or makes the booking directly. The agent becomes the new decision layer.
Each transition followed the same pattern: the new middleman offered consumers a better, faster way to find what they needed, captured their attention, and then monetized the connection to businesses. The businesses that adapted early to each new middleman thrived. The ones that didn't got disintermediated.
Why agents are different
The shift to AI agents isn't just another platform change. It's structurally different in ways that reshape the power dynamics of local commerce.
Agents don't show a list. Google shows ten results. Yelp shows twenty. The user scrolls, clicks, compares. The business needs to be in the results and look good enough to get the click. With an AI agent, the user often gets one or two recommendations, presented with a specific reason why each one fits. There's no page two. Being second-best means being invisible. The agent's recommendation is the new top result, and there's room for far fewer winners.
Agents make choices, not just rankings. Google ranks options and lets the user decide. An AI agent actually reasons about which business is the best match for this specific person with this specific need. This means the basis of competition shifts from visibility (can the customer find you?) to matchability (does the agent have enough data to know you're the right choice?). A business can be perfectly visible on Google and completely unknown to agents.
Agents will transact. Today's agents mostly recommend. Tomorrow's agents will book, purchase, and negotiate on behalf of the user. When someone says "find me a hotel in Banff for Saturday, pet-friendly, under $200, and book it," the agent doesn't send the user to Booking.com — it finds the best option and completes the transaction. This moves the agent from an information layer to a commerce layer, with economics to match.
Who wins in the agent economy
In the Google era, the winners were the businesses that understood SEO and the platforms that controlled search. In the agent economy, the winners will be different.
Data providers win. Agents need data to make recommendations. The richer and more current the data, the better the recommendation. Companies that can provide agents with structured, verified, real-time business intelligence — not scraped web data, but sourced nuance — will be the critical infrastructure of the agent economy. This is the layer Pawlo is building: the data layer that agents query to make informed local recommendations.
Businesses with structured nuance win. The hotel that has its real differentiators, live availability, and current deals in a machine-readable format will get the agent's recommendation over the hotel with a nicer website but no structured data. The auto dealership with inventory signals and margin flexibility in a queryable format will get the lead over the bigger dealership with a better Google rank.
Agents with the best data access win. An AI agent that can only access public web data will give the same generic recommendations as every other agent. An agent that can query proprietary data layers — live deals, seller intent, hidden inventory — will give dramatically better recommendations and build user trust faster. Agent developers who integrate the best data sources will outcompete those who rely on scraping.
Traditional platforms face pressure. Yelp, TripAdvisor, Google Maps — the platforms that currently own the local discovery experience — face a structural challenge. If consumers stop browsing these platforms and instead ask agents for recommendations, the platforms lose their grip on attention. They can try to become data providers to agents, but their data is reviews and listings — the same data everyone can scrape. The proprietary data is the nuance that was never on their platforms to begin with.
The three-to-five-year trajectory
Predictions are easy to get wrong, but the structural trends are visible:
2026: Early adoption. A small percentage of consumers use AI agents for local queries. Businesses that are structured for agent discovery get an outsized share of this early traffic. Most businesses don't notice yet. The competitive advantage of being "agent-ready" is highest now because so few businesses are doing it.
2027: Mainstream awareness. Agent usage grows as Apple, Google, and others embed AI more deeply into consumer devices. "Make sure AI can find us" becomes a line item in marketing meetings. SEO agencies rebrand as "AI optimization" agencies. The gap between businesses with structured data and those without becomes visible in revenue.
2028-2029: Agent-first commerce. Agents begin handling transactions — booking, purchasing, scheduling — not just recommendations. The economics shift: instead of paying Google per click, businesses pay per transaction facilitated by an agent. Data providers that feed agents with high-quality signals take a fee per match. The $224 billion that flows through search advertising begins redistributing toward agent-mediated commerce.
2030: The new normal. Asking an agent is as natural as Googling. Local businesses that aren't accessible to agents are the equivalent of businesses without a website in 2010 — technically operational but increasingly invisible. The data layer between businesses and agents is mature, standardized, and essential infrastructure.
What this means right now
If you build AI agents: your data sources are your competitive advantage. The model matters, but models are commoditizing. The data that feeds your agent's recommendations is what determines whether users trust you. Integrate data layers that give you access to information no other agent has — live signals, sourced nuance, seller intent.
If you run a local business: the window to get ahead of your competitors is open right now and it will close. The businesses that structure their nuance for AI agents in 2026 will be the default recommendations when agent-mediated commerce hits mainstream adoption. Every month you wait is a month your early-moving competitor builds a data advantage.
If you invest in or advise local businesses: the agent economy is the next platform shift. The economics are still forming, but the structural trajectory is clear. The middleman is changing again. The businesses and data layers that position themselves for agent-mediated commerce now will capture disproportionate value as the shift accelerates.
The Yellow Pages lasted forty years. Google's dominance has lasted twenty-five and counting. The agent economy is just starting. The businesses, builders, and data providers that recognize this early will define how local commerce works for the next decade.