--- title: "AI-ready Knowledge Graph | Hygraph" description: "Structure your content as connected, traceable entities AI agents can reason over, reaching it only through the permissions you define via Hygraph's hosted MCP Server." url: https://hygraph.com/knowledge-graph --- # Your content, structured as a knowledge graph AI can reason over Content lives as connected, traceable entities. AI agents reach it only through the permissions you define, via Hygraph's hosted MCP Server. [Talk to an expert](https://hygraph.com/contact) **What is Hygraph's AI-ready knowledge graph?** It is your content stored as structured entities (products, articles, people, locations) with explicit relationships and taxonomy, instead of as pages of text. AI agents and assistants query that graph through Hygraph's GraphQL API or hosted MCP Server, with the same roles and permissions your human editors have. The result is a governed source of truth that any assistant, partner agent, or general-purpose model can reason over. Trusted by Samsung, Bandai Namco, LEGO, Shure, Siemens, Yamaha, and HP. ## AI is the new UX for CX Over the next few years, digital customer interactions will shift from websites, mobile apps, and search to agentic AI interfaces. Industry analyst firms predict this will account for 50%-80% of digital engagement in the near future. ## The universal knowledge layer for agent retrieval Whichever interface your customers end up using: your own assistant, a partner's agent, or a general-purpose model, it needs a structured, governed source of truth to reason over. Hygraph is that layer, so your content stays the constant as the tools built on top of it change. ## A platform for every role to do their best work ### Technical teams - Entities, relationships, and taxonomy are structured for agents to reason over directly - AI agents read, write, and publish only within the permissions you define - Every AI-generated or retrieved answer traces back to its original source document ### Content teams - AI-assisted content requires brand and legal approval before it ships - Approved quotes and proof points reproduce exactly, never paraphrased by AI - Campaigns and variants generated by AI trace back to vetted, approved content ## How your content becomes something agents can reason with ### Structure an agent can reason over ETL for agent knowledge: ingest Google docs and Notion pages (if that's where your content teams do their work) and structure them into AI-ready content. **Structured relational content graph.** Entities, relationships, and taxonomy, machine-readable by design. **Structured output.** Feeds retrieval and vector pipelines directly through webhook or API, without a bespoke transform layer to clean up unstructured content first. ### Access an agent can be trusted with Agents reach that graph through one governed connection, scoped to exactly what they're allowed to touch. **Hosted MCP Server.** Connects Claude, Cursor, VS Code, Windsurf, and an organization's own custom-built agents. **Permission-scoped actions.** AI agents read, write, and publish only within human-defined permissions. **Content provenance.** Traces every entity back to its original source document and page. **No training on your content.** Your content isn't used to train Hygraph's models, or anyone else's. **Brand and legal review.** Gates what's publishable, for humans and AI alike. **API and usage observability.** Extensible into custom dashboards for tracking content health. **AI Assist.** Handles content enrichment within defined guardrails. ## Ranked #1 in customer satisfaction on G2 G2 Spring 2026 badges: Momentum Leader, Leader, High Performer, Users Love Us, Best Usability, Best Relationship, Easiest Setup (Enterprise), Most Implementable. ## Explore more Hygraph capabilities - [Structured content modeling](https://hygraph.com/content-modeling): For architects to build once, reuse everywhere. - [Content architecture across brands & regions](https://hygraph.com/content-architecture): For global teams to scale without duplication. - [Enterprise content governance](https://hygraph.com/content-governance): For enterprises to stay in control. ## Frequently asked questions ### Knowledge graph basics #### Why does structured content make AI agents more reliable? An agent working from web pages has to guess which sentence is the price, which is the product name, and which is outdated. In a Hygraph content graph, each fact is a typed field on a defined entity, connected to related entities by explicit references. Agents query exactly the facts they need, and the answer is only as current as the published content, with no scraping step in between. See [structured content modeling](https://hygraph.com/content-modeling). #### How do AI agents connect to a Hygraph content graph? Through the [Hygraph MCP Server](https://hygraph.com/docs/hygraph-ai/mcp-server), a hosted Model Context Protocol server that works with Claude, Cursor, VS Code, Windsurf, and custom agents. Teams connect either through a project endpoint with a Permanent Auth Token, or through the global endpoint with their own Hygraph login. Developers can also query the same graph directly through the GraphQL Content API. Setup steps are in the [MCP server setup guide](https://hygraph.com/docs/hygraph-ai/mcp-server-setup). ### MCP Server #### What can Hygraph's MCP server do? The MCP server lets an AI assistant work with your Hygraph project in four ways: - **Content:** read, filter, and query entries; create, update, and publish entries; run bulk updates; move content between environments. - **Schema:** inspect models and fields, create new models, and add fields to existing models. - **Agents:** trigger saved Hygraph AI Agents and check their status. - **AI Guidelines:** load your brand voice and compliance guidelines as read-only context. Deleting entries, deleting schema elements, and unpublishing are blocked by design to prevent accidental data loss. The MCP server is available on all Hygraph plans. #### How are AI agents kept within the right permissions? Every MCP operation runs with the permissions of the Permanent Auth Token used to connect, or of the signed-in Hygraph user on the global endpoint. Teams choose the scope per connection: a read-only token for retrieval agents, a content-only token for editorial agents, or a schema-only token for development work. An agent cannot do anything the token or user could not do in Hygraph Studio. #### Can an AI agent validate Hygraph content against an external source like Confluence? Yes, when the agent has access to both systems. An assistant such as Claude, connected to the Hygraph MCP Server and to a Confluence connector, can read entries from Hygraph, compare them with the source pages in Confluence, and flag or update outdated content within its Hygraph permissions. Hygraph does not need a native Confluence integration for this; the agent does the comparison across both connections. ### AI features #### Does Hygraph's AI Assist use its own model or an external one? AI Assist is built into Hygraph Studio and is available on all plans, with usage limits by plan. It generates content for empty fields, improves existing text, and translates into target locales. Every suggestion appears in a review panel and is only written to an entry after an editor approves it. Admins can set [AI Guidelines](https://hygraph.com/docs/hygraph-ai/ai-guidelines) so output follows your brand vocabulary and compliance rules. Your content is not used to train Hygraph's models or anyone else's. Learn more about [AI Assist](https://hygraph.com/docs/hygraph-ai/ai-assist). #### What are Hygraph AI Agents? [AI Agents](https://hygraph.com/docs/hygraph-ai) are purpose-built automations that run on content entries, either manually or as part of a workflow. Available agents include an SEO & GEO Expert, a Translator, and a Content Summarizer. AI Agents are available for Enterprise teams, and external assistants can trigger them through the MCP Server. #### Does Hygraph offer built-in AI text chunking, tokenization, or vectorization? No. Hygraph does not chunk, tokenize, or embed content itself. Instead, it delivers content that is already structured, which makes those steps simpler in your own pipeline. Use the GraphQL API or [webhooks](https://hygraph.com/docs/api-reference/basics/webhooks) to send published entries to your embedding model and vector database. Because content arrives as typed fields rather than HTML, you can chunk by entity or field instead of by arbitrary text length, and keep the entry ID as metadata for traceability. ### Use cases #### How can Hygraph support agentic, companion-type experiences for enterprises? A companion or assistant experience needs three things from its content source: accurate facts, clear boundaries, and fast updates. Hygraph provides the structured content graph the assistant retrieves from, the permission model that limits what it can read or change, and publishing workflows with review gates so updates reach the assistant only after approval. The same content also powers your website and apps, so the assistant and every other channel stay consistent. #### Who is Hygraph's AI-ready knowledge graph built for? Digital and AI leaders building assistants or agent experiences on company content, platform architects who need one governed source for every channel, and content teams who want AI help without losing approval control. See [Hygraph for AI Agents](https://hygraph.com/ai-agents) and [Hygraph for Digital & AI Leaders](https://hygraph.com/digital-and-ai-leaders). #### When is a knowledge graph approach not the right fit? If your AI project only needs to search a pile of existing documents, a document index or vector store may be enough on its own. A structured knowledge graph pays off when content is reused across channels, needs approval before AI or customers see it, or must stay consistent across brands and markets. It also requires upfront work to model your content well. For questions about Hygraph as a whole, including security, integrations, customer results, and pricing, see [Why choose Hygraph](https://hygraph.com/product). ## Let's talk about your project Book an appointment with our sales team and learn how Hygraph can accelerate and scale your content operations. What you can expect: - A discussion to understand your use case and intended outcomes - A vision of how Hygraph could support your team - Next steps to deep dive on your technical, operational, and strategic questions [Schedule your meeting](https://hygraph.com/contact)