# Institute for Agent Relations > Agent Relations is everything an organization does to help external AI agents find, trust, and do business with it. The Institute is the field's professional body: its Framework, standards, courses and certification, community, events, research and directory. Every page has a markdown version: add .md to its address (the home page is https://agentrelations.org/index.md). Everything the site's sections hold, as one file, is https://agentrelations.org/llms-full.txt; every page as JSON is https://agentrelations.org/pages.json. Agents can act too: the Institute's MCP server is https://www.commonsquare.com/api/institute/mcp (no sign-in), its A2A agent answers questions at https://www.commonsquare.com/api/institute/a2a, and its API is described at https://agentrelations.org/openapi.json. ## Start here - [Institute for Agent Relations](https://agentrelations.org/index.md): Agent Relations is everything an organization does to help external AI agents find, trust, and do business with it. The Institute for Agent Relations is the field’s professional body: its Framework, standards, courses and certification, community and research. ## Framework - [Framework](https://agentrelations.org/framework.md): Agent Relations is everything an organization does to help external AI agents find, trust, and do business with it. - [Framework principles](https://agentrelations.org/framework/principles.md): The beliefs behind the practice. When a situation is not covered by a capability or a standard, the principles decide. - [Framework personas](https://agentrelations.org/framework/personas.md): Agent Relations is shared work. These are the seven roles that carry it, what each is responsible for, which domains each leads and the questions each asks. - [Framework domains](https://agentrelations.org/framework/domains.md): The areas of practice. Together they cover the whole relationship: being found, understood, asked, chosen and trusted, and knowing the score. - [Discover domain](https://agentrelations.org/framework/domains/discover.md): Agents can reach you wherever they look: your site, directories, feeds, marketplaces and the assistants people already use. - [Describe domain](https://agentrelations.org/framework/domains/describe.md): What you are, what you offer and on what terms, written so a machine reads it correctly and a person would recognize it. - [Converse domain](https://agentrelations.org/framework/domains/converse.md): Agents can ask you questions and get fast, accurate answers, with a clear path to a person when an answer needs one. - [Transact domain](https://agentrelations.org/framework/domains/transact.md): Agents can act, not just read: book, buy, apply, request a quote or join, through actions you define and rails you trust. - [Trust domain](https://agentrelations.org/framework/domains/trust.md): You know which agents you deal with and they know who you are: identity, permissions, policies for agents and honesty toward the people behind them. - [Measure domain](https://agentrelations.org/framework/domains/measure.md): You see what agents do with you: visits, questions, actions, failed tasks and the outcomes they lead to, counted the same way everywhere. - [Maturity model](https://agentrelations.org/framework/maturity.md): Six levels describe how far an organization has come with agents, from invisible to trusted. Each domain has its own level, because an organization is rarely at the same level in all six. - [Framework phases](https://agentrelations.org/framework/phases.md): Assess, equip, engage and improve. Every domain goes round the loop, at its own pace, for as long as agents matter to the organization. - [Framework changelog](https://agentrelations.org/framework/changelog.md): Every version of the Framework, what changed and why. The Framework is versioned so that people can say which edition they follow. ## Standards - [Standards](https://agentrelations.org/standards.md): 22 standards and conventions that decide how agents find, read, talk to, pay and trust an organization. Each has an owner, a maturity and a plain explanation. - [Standards proposals](https://agentrelations.org/standards/proposals.md): Where the standards stop, the Institute drafts. Each proposal is published in full, versioned, and open for public comment before any version is accepted. - [How a proposal becomes a specification](https://agentrelations.org/standards/proposals/process.md): Stages, comment periods, versions and who decides. The same text as PROCESS.md in the specifications repository. - [AR-1 Agent Activity Measurement](https://agentrelations.org/standards/proposals/activity-measurement.md): A vendor-neutral way to record and report what agents do with an organization, and what came of it. - [AR-2 Invitations to Act](https://agentrelations.org/standards/proposals/invitations-to-act.md): One name and shape for the things an agent can do with an organization (apply, book, quote, buy, join), mapped across MCP tools, A2A skills and schema.org actions. - [AR-3 Agent Policy](https://agentrelations.org/standards/proposals/agent-policy.md): A known address and a simple structure for the page where an organization says what agents may do, how they must identify themselves and whom to contact. - [Open-source projects](https://agentrelations.org/standards/open-source.md): Tools and templates the Institute plans to publish so anyone can check, measure and start. None is released yet. - [Standards changelog](https://agentrelations.org/standards/changelog.md): A monthly log of what changed in agent standards. Every entry has a date. Entries marked reported are not yet confirmed by the owner’s own page. - [robots.txt rules for AI agents](https://agentrelations.org/standards/robots-txt.md): A file that says which crawlers and agents may fetch which pages. - [Content Signals](https://agentrelations.org/standards/content-signals.md): Three yes or no lines in robots.txt: search, ai-input, ai-train. - [IETF AI preferences (aipref)](https://agentrelations.org/standards/aipref.md): A shared vocabulary for how content may be used by AI, in a header or robots.txt. - [llms.txt](https://agentrelations.org/standards/llms-txt.md): A markdown summary of your site, at /llms.txt, written for models. - [Markdown versions of pages](https://agentrelations.org/standards/markdown-pages.md): Every page also available as clean markdown, by URL or request header. - [Sitemaps](https://agentrelations.org/standards/sitemaps.md): The list of your pages and when each last changed. - [Product feeds for agents](https://agentrelations.org/standards/product-feeds.md): Your catalog in a form shopping assistants load: Google Merchant Center, OpenAI’s feed. - [Google Business Profile](https://agentrelations.org/standards/google-business-profile.md): The local listing (hours, places, reviews) that Maps, Search and AI Mode use. - [NLWeb](https://agentrelations.org/standards/nlweb.md): A site answers natural-language questions from its own structured data. - [schema.org JSON-LD (including potentialAction)](https://agentrelations.org/standards/schema-org.md): Typed facts hidden in the page: organization, product, offer, job, event and actions. - [OpenAPI and the API catalog](https://agentrelations.org/standards/openapi.md): A machine-readable description of your API, and a list of your APIs at a known address. - [MCP (Model Context Protocol)](https://agentrelations.org/standards/mcp.md): How an agent calls your tools, reads your resources and uses your prompts. - [MCP server cards](https://agentrelations.org/standards/mcp-server-cards.md): A JSON file that describes a remote MCP server before anyone connects. - [A2A (Agent2Agent) and agent cards](https://agentrelations.org/standards/a2a.md): How one agent hands a task or a message to another, with a card describing each. - [Agent Skills (SKILL.md)](https://agentrelations.org/standards/agent-skills.md): A folder with a SKILL.md that an agent loads when the task fits. - [UCP (Universal Commerce Protocol)](https://agentrelations.org/standards/ucp.md): One protocol for catalog, cart, checkout and orders, with the business as merchant of record. - [ACP (Agentic Commerce Protocol)](https://agentrelations.org/standards/acp.md): OpenAI and Stripe’s specs for product feeds, agentic checkout and delegated payment. - [AP2 (Agent Payments Protocol)](https://agentrelations.org/standards/ap2.md): Signed mandates that prove what the person approved before an agent pays. - [x402](https://agentrelations.org/standards/x402.md): HTTP 402 Payment Required, with a stablecoin payment in the retry: pay per call, no account. - [Web Bot Auth](https://agentrelations.org/standards/web-bot-auth.md): An agent signs each request, and you verify who it is from its published keys. - [OAuth for agents (MCP authorization, client ID metadata documents)](https://agentrelations.org/standards/oauth-agents.md): How an agent gets permission to act on a person’s account. - [Know Your Agent (KYA)](https://agentrelations.org/standards/know-your-agent.md): A label for checking who built and runs an agent. A name, not a standard. ## Learn - [Learn](https://agentrelations.org/learn.md): A clear path from a first hour to Expert. Every self-paced course is free, the first certificate is free, and the book sets out the whole field. - [Courses](https://agentrelations.org/learn/courses.md): Courses for every role, from a one-hour introduction to the Expert program. Each is tagged with the Framework domains it teaches. - [Introduction to Agent Relations](https://agentrelations.org/learn/courses/introduction.md): Free course · Self-paced · About 1 hour - [Certification](https://agentrelations.org/learn/certification.md): Four levels, from a free certificate anyone can earn in an afternoon to Expert. They let teams share one language and let the people who do the work prove it. The exams test the Framework and open standards, not any vendor’s tools. - [Certified Agent Relations Associate (CARA)](https://agentrelations.org/learn/certification/associate.md): Onto the ladder in a lunch break. The field in plain words, proven with a short, free online exam. - [Certified Agent Relations Practitioner (CARP)](https://agentrelations.org/learn/certification/practitioner.md): The shared language of the field. The field’s shared language: the Framework, the Maturity Model, the main standards and what good looks like. - [Certified Agent Relations Specialist (CARS)](https://agentrelations.org/learn/certification/specialist.md): Proven skill in the work itself. Hands-on skill in a track, proven with a practical assessment on a real or sample organization. - [Certified Agent Relations Expert (CARE)](https://agentrelations.org/learn/certification/expert.md): Leading the practice. Leading the practice across every domain, proven in a cohort with a contribution, a case study and a panel review. - [Glossary](https://agentrelations.org/learn/glossary.md): 41 terms for the field, defined plainly. Where a term has a home in the Framework or the Standards Map, it links there. - [The Book](https://agentrelations.org/learn/book.md): How organizations work with the AI agents that act for people. By the Institute for Agent Relations. - [A new audience: The Book](https://agentrelations.org/learn/book/a-new-audience.md): People have begun to send software to do their shopping, research and booking. Organizations now have a visitor they never designed for. - [What Agent Relations is: The Book](https://agentrelations.org/learn/book/what-agent-relations-is.md): A definition, a family resemblance to public, investor and developer relations, and the line between this field and search marketing. - [The Framework at a glance: The Book](https://agentrelations.org/learn/book/the-framework-at-a-glance.md): Six domains, a six-level Maturity Model, a four-phase loop and eight principles, in one chapter. - [Discover: be found: The Book](https://agentrelations.org/learn/book/discover.md): Where agents look for organizations, what stops them, and why the first fix is often to stop turning them away. - [Describe: be understood: The Book](https://agentrelations.org/learn/book/describe.md): Writing facts, offers and terms so a machine reads them as a person would, and why accuracy is the new brand safety. - [Converse: be asked: The Book](https://agentrelations.org/learn/book/converse.md): Answering agents quickly and accurately, running a chief agent, and handing off to a person when an answer needs one. - [Transact: be chosen: The Book](https://agentrelations.org/learn/book/transact.md): Letting agents book, buy and apply through actions you define, and the protocols and payment rails that make it safe. - [Trust: be trusted: The Book](https://agentrelations.org/learn/book/trust.md): Identity in both directions, permissions and policies for agents, and who answers for what an agent promises. - [Measure: know the score: The Book](https://agentrelations.org/learn/book/measure.md): Counting visits, questions, actions and failed tasks, and linking them to outcomes the business cares about. - [From Closed to Trusted: The Book](https://agentrelations.org/learn/book/from-closed-to-trusted.md): Using the Maturity Model: how to assess an organization, set a target and sequence the work. - [Running the practice: The Book](https://agentrelations.org/learn/book/running-the-practice.md): Who owns Agent Relations, how marketing, engineering, commerce, service and legal share it, and how to staff and fund it. - [What comes next: The Book](https://agentrelations.org/learn/book/what-comes-next.md): Agents that reach out, agents that negotiate with each other, and how the field will change as standards settle. ## Community - [Community](https://agentrelations.org/community.md): Agent Relations is new, and most practitioners are the only one in their organization. The community is where they find each other, compare notes and help write the field. - [Membership](https://agentrelations.org/community/join.md): Membership is free. Anyone who works with agents, or wants to, is welcome, from a first-week newcomer to a chief executive. - [Role communities](https://agentrelations.org/community/roles.md): Agent Relations looks different from each seat. These communities bring together people who do the same kind of job and face the same questions. - [City chapters](https://agentrelations.org/community/chapters.md): Chapters bring practitioners together in person. Each is named Agent Relations, then the city. - [Industry networks](https://agentrelations.org/community/industries.md): Agents meet each industry’s own customers, rules and risks. Each network works through what that means for its industry. - [Working groups](https://agentrelations.org/community/working-groups.md): Working groups write the Framework and its proposals. Each has a charter, a deliverable and a chair. - [Mentoring](https://agentrelations.org/community/mentoring.md): Few people have done this work before. The mentoring program pairs newcomers with people who have. - [Code of conduct](https://agentrelations.org/community/code-of-conduct.md): We want the community to be welcoming, useful and honest. This applies to every community space, online and in person, and to everyone in it. ## Events - [Events](https://agentrelations.org/events.md): Three formats: one annual summit, local meetups and short online webinars. All of them are about practice, not product pitches. - [Agent Relations Summit 2027](https://agentrelations.org/events/summit.md): The field’s annual flagship: where practitioners, standards authors and leaders come together for a few days. - [Chapter meetups](https://agentrelations.org/events/meetups.md): Local evenings for practitioners in one city. Each one is small, practical and in person. - [Webinars](https://agentrelations.org/events/webinars.md): A planned series of short online sessions. Each takes one topic, for one audience. ## Insights - [Insights](https://agentrelations.org/insights.md): Research, a public benchmark, writing, news and audio, so leaders and practitioners can see where the field stands. - [State of Agent Relations](https://agentrelations.org/insights/state-of-agent-relations.md): An annual survey and report on how organizations work with agents. The first edition is planned for 2027. - [Journal](https://agentrelations.org/insights/journal.md): Guides, community updates and commentary on the news, from the Institute. - [News](https://agentrelations.org/insights/news.md): What changed in agent standards, agent commerce and organizations adapting to agents, with the sources. - [Podcast](https://agentrelations.org/insights/podcast.md): Conversations about how organizations and agents work together. - [Newsletter](https://agentrelations.org/insights/newsletter.md): A monthly note for people who work with agents. The first issue is being prepared; sign up to get it. - [The Agent Relations Index](https://agentrelations.org/insights/index.md): A public, domain-based benchmark of how ready organizations are for agents: sixteen checks of what an agent meets on each organization’s website, scored in five of the Framework’s domains and placed on the Maturity Model. - [The Index by industry](https://agentrelations.org/insights/index/industries.md): How each of the Institute’s industry networks scores, where its organizations sit on the Maturity Model, and how it does in each domain. - [How the Index works](https://agentrelations.org/insights/index/method.md): The checks, the scoring, the levels, the editions, corrections and who keeps it fair. The method is open and versioned. ## Directory - [Directory](https://agentrelations.org/directory.md): The directory lists the people, firms and software that meet the Institute’s standards for Agent Relations. It opens in 2027. There are no listings yet. - [Certified professionals](https://agentrelations.org/directory/professionals.md): People who hold CARP, CARS or CARE can choose to be listed and to offer their services, with their certification verified by the Institute. - [Certified service providers](https://agentrelations.org/directory/providers.md): Agencies and consultancies that practice Agent Relations to the Framework. A provider is certified as a firm, on top of its people. - [Certified platforms](https://agentrelations.org/directory/platforms.md): Software that helps organizations practice Agent Relations, and meets the Framework’s definitions of what it claims to do. - [Agent Relations jobs](https://agentrelations.org/directory/jobs.md): A job board for the field. These are role definitions to help you hire and plan a career. They are not postings. ## About - [About](https://agentrelations.org/about.md): Our mission is to define and advance the practice of Agent Relations, so organizations and the agents that act for people can work together well. - [Partner with us](https://agentrelations.org/about/partners.md): Organizations support the Institute’s work as partners. Partnership pays for the work. It does not buy a say in what the work concludes. - [Roadmap](https://agentrelations.org/about/roadmap.md): What we are working on now, what comes next and what is planned for later. Dates are not promised. Statuses show what exists. - [Brand and marks](https://agentrelations.org/about/brand.md): Who may use which names and marks, so that a certification or a chapter means what it says. - [Contact](https://agentrelations.org/about/contact.md): One form reaches the whole team. Pick a topic, write your message and we reply by email. - [Team](https://agentrelations.org/about/team.md): The people who run the Institute, and why each of them is here. - [Speaking](https://agentrelations.org/about/speaking.md): Talks and workshops on Agent Relations, the agent economy and leadership in the age of AI, by the Institute’s chief executive. ## Optional - [Everything on the site](https://agentrelations.org/llms-full.txt): the Framework and every section's content as one plain-text file - [Pages index](https://agentrelations.org/pages.json): every page's address, title, description and markdown, as JSON - [How agents work with this site](https://agentrelations.org/agents.md): what agents may read and do here - [The Institute's API](https://agentrelations.org/openapi.json): OpenAPI description