The Side Door — When AI Agents Check Into Your Polymath Website

Part 5 of the Polymath Website Series

by David Anttony

November 1880. Monte Carlo.

César Ritz stepped out of a carriage into the Mediterranean dusk. He was thirty years old. The youngest of thirteen children from a Swiss mountain village. Everything he owned fit in a leather case — a pressed suit and a letter of appointment from Colonel Pfyffer d’Altishofen, the man who had been tracking his career in a private scrapbook for years before offering him the position. That letter meant more to Ritz than the job itself. It meant someone of standing believed the farm boy from Niederwald belonged in rooms like these. He had read it more than once on the fourteen-hour train from Lucerne.

He stood in the forecourt of the Grand Hôtel and didn’t move.

Across the square, Charles Garnier’s opera house glowed — the same architect behind the Paris Opéra. To the left, the Casino de Monte-Carlo hummed with money in motion, crystal chandeliers over roulette tables where Europe’s aristocracy placed bets they could afford to lose. Hotels, gardens, restaurants, cafés — all connected. All one system. Twenty years earlier, this plateau had been barren rock. François Blanc had built the entire ecosystem from nothing.

Ritz didn’t know it yet — couldn’t possibly have known — but everything that would define his life was already inside that building. The chef he would partner with for thirty years. The owner’s twelve-year-old niece, somewhere in the upper floors, who would one day become his wife. And a vision of how different doors serve different visitors inside one seamless machine — a vision that would stamp his name in the stone of history. Ritz. A single word that means quality, focus, and the relentless pursuit of excellence. From that single vision, Ritz built the Hôtel Ritz Paris, the Ritz London, the Carlton in London, and a hospitality philosophy that defined luxury for a century. His wife Marie-Louise ran the Paris Ritz for decades after his death. His name became the standard. And 118 years after he stepped out of that carriage in Monte Carlo, the Ritz-Carlton brand was acquired by Marriott International for approximately $500 million. Half a billion dollars — for a name that started with a farm boy, a suit, and a letter.

He walked in thinking it was a winter job. It was the beginning of everything.

The Polymath Website Series

A polymath is a person whose knowledge spans multiple disciplines — someone who solves problems that specialists cannot by connecting completely different fields. César Ritz was exactly that. He combined guest psychology, military discipline, kitchen engineering, architecture, brand-building, and theatrical instinct into one seamless system. He was a polymath without ever being called one.

At Seresa, we coined the term Polymath Website™ because that’s exactly what the next generation of website must be: a single integrated system that knows everything about your business and exactly how to serve whoever — or whatever — walks through the door.

This is the fifth instalment of the Polymath Website series:

In Parts 1 through 4, we built a Polymath Website with six layers and one door — the front door, where humans arrive.

In Part 5, the stack grows. Because a new kind of visitor has arrived. Not a human. An AI agent. And it needs its own entrance.

Ritz Built Three Doors

What Ritz saw in Monte Carlo — what stopped him in that forecourt — was a system with three distinct surfaces that guests experienced as one seamless whole.

Three doors. One hotel. One brand.

Your Polymath Website needs the same architecture. And in October 2026, it needs a door that Ritz never imagined.

Layer 7: The Side Door — Agent-to-Agent API

A new kind of visitor is arriving at your website. Not a human. An agent — an AI shopping assistant acting on behalf of a human. Researching, comparing, negotiating, and in some cases buying — without the human ever opening a browser.

This agent doesn’t need your homepage. Doesn’t need Cherry Rose to greet it. It needs a side door — a functional entrance. Pure API. Pure JSON. No UI.

Behind that door, there are different rooms for different interactions:

Every interaction is 100% API-driven. 100% JSON. Zero UI.

The front door is warm. The side door is ruthlessly efficient. Both are essential.

This Isn’t a Prediction. It’s Happening Right Now.

The infrastructure is being built this year. By the biggest names in technology and finance:

Gartner forecasts that 40% of enterprise applications will embed task-specific AI agents by the end of 2026 — up from fewer than 5% in 2025.

The side door isn’t something you plan for in 2028. You need to build it now.

You Don’t Need a Protocol. You Need a Sign.

The tech industry is fighting over which discovery protocol wins — agents.txt, agent-card.json, mcp.json, ai-catalog.json, llms.txt, openapi.json. At least seven competing standards. All drafts. None adopted at scale. Research from Vercel shows that 99% of the top 100 websites fail basic content negotiation with AI agents.

We think that’s the wrong approach.

If you’re running a high-quality hotel, you don’t want delivery trucks at the front door. You’ve solved it simply — a clear sign at the service entrance, a directory board behind it telling every visitor which room to go to. You don’t build six separate buildings. You build one service entrance with different rooms behind it.

One API. One sign. Different query points inside.

Every AI agent already reads robots.txt first. So you put the sign where everyone’s already looking — and you make sure every obvious probe leads to the same room:

When an agent hits the root endpoint, it gets the directory board:

{
  "name": "Your Business",
  "endpoints": {
    "overview":   "/wp-json/agent/v1/",
    "identity":   "/wp-json/agent/v1/identity",
    "catalogue":  "/wp-json/agent/v1/catalogue",
    "pricing":    "/wp-json/agent/v1/pricing",
    "transact":   "/wp-json/agent/v1/transact",
    "service":    "/wp-json/agent/v1/service"
  },
  "concierge":  "/wp-json/agent/v1/ask"
}

/identity — who you are and what you do. /catalogue — what you sell, structured. /pricing — how much, what tiers. /transact — how to buy. /service — post-purchase relationship. /ask — the concierge fallback, where an agent sends a natural language query and gets directed.

One API. Six query points. One concierge.

Different AIs will start at different query points. It doesn’t matter. Every query point reaches the same system. Different questions. Same hotel.

And here’s what makes this moment unprecedented: user activity is driving the protocols — not the other way around. HTTP was defined before the web existed. But right now, AI agents are already shopping, already holding credit cards, already completing transactions — and the standards bodies are scrambling to catch up. The agents.txt IETF draft was published in June 2026, months after Mastercard was processing live agent transactions.

The cars are driving through the fields. The businesses that put up signs now will be the ones the traffic finds.

The Intelligence Layer

Above all of this — above the front door, the side door, and the back kitchen — sits the intelligence layer. The review. The daily analysis that tells you what’s happening before you ask.

This isn’t the data warehouse. Collection is infrastructure — plumbing. The intelligence layer is thinking. It can be built into the website itself. Or it can be outsourced — data goes out to an external service, comes back as a complete analysis package, like sending your accounts to a bookkeeper and getting a report every Monday.

Every day — automatically — the system runs analysis across your data. Patterns. Trends. Anomalies. Opportunities. Layered on top of yesterday’s insights. Compounding.

So when you sit down on a Monday morning and say “What should I improve this week?” — the answer is already prepared. The analysis was done at 3am. The recommendations are waiting.

You didn’t ask for a report. The report was already written.

This is Ritz himself. Not the doorman, not the chef, not the floor manager — the owner. The intelligence layer sitting above everything, seeing patterns before anyone else, solving problems before they become problems. He knew a guest was unhappy before the guest complained. He knew the kitchen was falling behind before the plates were late.

The Polymath Website gives that owner everything they need to see — so they can do what only a human can do. Think ahead.

The Updated Stack: 8 Layers, 3 Doors

Layers 1–5 serve humans. Layer 6 serves the system. Layer 7 serves agents. Layer 8 serves the owner.

The guest experiences delight. The kitchen experiences discipline. The service corridors experience efficiency. The owner sees everything.

The Dirty Old Truck

If you’ve read this far, you might be thinking: this is fascinating, but it feels remote. Agent-to-agent commerce. Tokenised credit cards for AI. Protocol wars. Side doors. It sounds like something for the big players — not for me and my business.

But here’s the thing.

You probably noticed we’ve used WordPress examples throughout this article. The REST API at /wp-json/. The agent namespace. The WooCommerce endpoints. The robots.txt signage.

That’s not an accident. And that’s for a very good reason.

In the original 1984 Transformers series, the leader of the Autobots — the most powerful Transformer of them all — was disguised as a beat-up, flat-nose Freightliner cab-over truck. Dusty. Ordinary. The kind of vehicle you’d drive past without a second glance.

His name was Optimus Prime.

On August 26, 2026, Peter Cullen — the man who voiced Optimus Prime for four decades — passed away at the age of 85. And something interesting happened. The Transformers had always been an allegory — a fictional story about machines that are more than they appear. But Cullen’s death put a human face on the myth. It made the story real. The man behind the voice was gone, and suddenly the character meant something deeper than fiction.

WordPress is Optimus Prime.

It looks like a dirty old truck. A blogging platform from 2003. The thing your aunt uses to post recipes. The CMS that developers love to mock at conferences.

But under the bonnet?

Every single piece of infrastructure the Polymath Website needs — the front door, the side door, the back door — WordPress already has it. Not as a bolt-on. Not as a plugin someone built last week. As core architecture that’s been running in production for over two decades.

The REST API is the side door. WooCommerce is the checkout layer. The plugin architecture is extensibility itself. And the 40% market share means more AI agents will encounter WordPress than any other platform on earth.

WordPress doesn’t need to become something new. It needs to transform — to reveal what’s already there.

Just like Optimus Prime. The power was always under the hood. It just looked like a truck. The only layer WordPress doesn’t have out of the box is Layer 8 — the intelligence layer. Everything else — the front door, the side door, the data pipeline, the API infrastructure — is already there. Waiting to be switched on. And here’s what’s changed: building plugins and custom functionality for WordPress has never been easier. AI-assisted development means that the gap between “I have an idea” and “I have a working plugin” has collapsed from months to days.

Start Building Your Polymath Website — Watch the Series

From reading about the future to building it — on WordPress, with tools you already have.

We’re publishing step-by-step instructional videos showing you how to build every layer of the Polymath Website on WordPress — right now, with the tools you already have.

Practical. No theory. Built for WordPress site owners who want to stop reading about the future and start building it.

The Hotel That Became a Brand

Ritz didn’t just build a hotel. He built a brand machine.

The Savoy wasn’t just accommodation. Ritz’s name became so synonymous with luxury that it entered the English language — ritzy. Escoffier’s kitchen created Peach Melba and Crêpes Suzette — products, branded, exported far beyond the hotel walls. A great hotel is a lot more than a place to sleep. It’s a brand manager. A product creator. A system of tremendous intelligence across every layer.

Your Polymath Website is that system. Your AI concierge. Your invisible kitchen brigade. Your side doors serving agents that bring customers while you sleep. Your intelligence layer compounding insights every day.

Ritz understood this in 1889: every interaction builds trust, reputation, and a brand that transcends the building.

Your Polymath Website is that building. It never closes. It never sleeps. And in 2027, it serves both humans and their AI agents — through whatever door they need.

Start Building Your Side Door Today

The agents won’t wait for your contact form.

Transmute Engine™ is the data layer that powers every door of your Polymath Website — front, back, and side. It collects, processes, and routes your data so every layer has the signal it needs.

No credit card. No commitment. Full access for 30 days.

Learn more about Transmute Engine™ →

P.S. The agent economy isn’t coming. It’s here. Visa, Mastercard, American Express, Google, and OpenAI have all launched agent commerce products this year. Your competitors’ AI agents are already shopping for solutions. The question is whether your website has a door they can walk through — or whether they’ll walk past and find one that does. Start planting your Data Trees → Every Side Door transaction your Polymath Website handles tomorrow depends on the data it collects today.

[ Polymath Website™ is a term coined by Seresa. We encourage its use by anyone in the industry. We simply ask that when referencing the concept, you credit Seresa as the originator. No permission needed. ]

This article was written by AI + David Anttony. The thinking is human. The typing had help.