# Agent-Ready Websites by StudioNN

Websites that remain excellent for people and become reliable, permission-aware tools for their agents.

## What changes

AI search visibility helps agents find and understand a company. Agent readiness adds structured capabilities and tools so an authorised agent can complete a valuable job.

## The five layers

### 01 Readable
Clear information architecture, semantic HTML, structured data and stable machine-readable documents.

Included outputs: Semantic HTML, Schema.org, llms.txt, Markdown.

### 02 Discoverable
Explicit capability manifests and agent documentation so an agent knows what exists and how to connect.

Included outputs: MCP manifest, Capability map, Agent guide, Tool schemas.

### 03 Operable
Typed tools connected to real business logic through WebMCP, remote MCP or existing APIs.

Included outputs: WebMCP, Remote MCP, API adapters, Live state.

### 04 Governed
Authentication, permissions, validation and explicit confirmation for actions that create cost or change state.

Included outputs: Session auth, Role scopes, Confirmations, Audit trail.

### 05 Measurable
Evaluations and instrumentation that show whether agents discover, complete and convert the intended jobs.

Included outputs: Task evals, Success rate, Failure logs, Agent attribution.

## Delivery process

01. **Find the valuable jobs.** We map the questions and actions where an agent can remove friction or create revenue.
02. **Design the contracts.** Each tool gets a precise purpose, inputs, outputs, permissions and failure behaviour.
03. **Connect the real system.** We build the machine-readable layer and connect tools to inventory, booking, CRM, product or account systems.
04. **Test the full journey.** We run real agent workflows, test edge cases and measure completion before calling the system ready.

## Audit areas

- Human and agent job map
- Machine readability and entity clarity
- Capability and tool discovery
- Backend actionability and live data
- Authentication, permissions and confirmations
- Task evaluations, observability and attribution

## Core principles

- No live data means no tool.
- No confirmed action means no mutation.
- Public information stays public. Private information stays inside the user's existing permissions.
- A real workflow is tested from discovery to completion before it is called ready.

## Connect

- Service page: https://studionnagency.com/agent-ready
- Agent guide: https://studionnagency.com/agents
- MCP endpoint: https://studionnagency.com/mcp
- Contact: felipe@studionnagency.com