If you follow the evolution of AI agents, you have probably heard about MCP (Model Context Protocol), the open standard that allows assistants such as Claude or ChatGPT to connect directly to a company’s tools and data, without custom integrations or fragile scripts. A document management system with MCP is precisely that: a document management system that exposes its data and functions through this protocol so that any AI can use them natively.
In the past, before MCP, we had already shown how we could connect ChatGPT—a custom GPT—to Athento. Remember? However, as you can see in the article, it was a lot of work! Integrating ChatGPT and Athento required a certain level of technical knowledge. Moreover, there was no way to do the same with other AIs such as Claude.ai.
Fortunately, technology is advancing incredibly fast, and standards now exist that make our work easier and help us avoid being tied to a specific AI.
At Athento, we have understood this for a long time: a document management system is only as useful as its ability to bring information to where people—and now agents—need it. That is why, in May 2026, we became the first document management system with MCP on the market, in production and fully operational. Since then, we have continued expanding its capabilities, and we have no intention of stopping, because our customers are bringing us increasingly exciting use cases.
In this article, we explain exactly what MCP is, what Athento’s MCP can already do, what possibilities it opens up for your company, and how you can connect it to Claude or ChatGPT today.
The Model Context Protocol is a standard that defines how an artificial intelligence model can “talk” to an external system: what actions it can request, what data it can access, and how it authenticates securely. Before MCP, connecting an AI assistant to a document management system meant building specific integrations for each model, maintaining them, and keeping them updated.
With MCP, Athento exposes a catalog of “tools” that any compatible client—Claude, ChatGPT, or an internal agent built by your team—can automatically discover and use. The practical result: a document management system with MCP is no longer a system that users need to open and consult manually. Instead, it becomes another capability that your AI agents can use on your behalf.
A document management system with MCP presents certain challenges that do not exist in vertical applications such as a CRM or ERP. Vertical applications have a limited scope and clearly defined entities. For example, a CRM works with concepts such as Accounts, Deals, Quotes, etc. In document management systems—or at least those that allow business processes to be implemented, such as Athento—the situation becomes more complex. Within the same document management system, different “entities” representing different processes may coexist: an invoice, an employee, a purchase order, or a meeting record.
The MCP must therefore be capable of understanding “whatever happens to live inside the document management system.” This is one of the reasons why building a document management system with MCP is more complex. Fortunately, our agile philosophy and our commitment to providing modern, cutting-edge solutions guided our product team in equipping Athento with the necessary tools so that any customer, regardless of their business case or use cases, can use Athento’s MCP without touching a single line of code.
Athento’s MCP server was introduced on May 20, 2026, with MCP server functionality integrated into the core of the platform, making Athento the first document management system with MCP available on the market. That same day, we also added the knowledge base search tool (Athento’s RAG functionality), and over the following days and weeks we continued strengthening the most delicate part of any integration of this kind: authentication.
That same month, we added OAuth discovery, and by the end of June we had refined the flow to make it compliant with the specification: OAuth metadata is exposed through the standard .well-known URLs, as required by the most demanding MCP clients, including Claude. It may sound like a technical detail, but it matters. The first version worked perfectly with ChatGPT, but getting Claude to connect took us a little longer 🙂.
The MCP design follows Athento’s product development philosophy: a lightweight product, rapid iteration, and a solid foundation on which to keep building. Other tools on the market are still testing their MCP implementations, but we were in a hurry to become users of Athento’s MCP ourselves, since we use our own platform to manage Athento’s commercial operations, DevOps, and product operations.
The server exposes a single endpoint that follows the JSON-RPC 2.0 standard and provides five tools:
| Tool | What it allows you to do |
|---|---|
search_documents | Search for documents by text within a space |
get_document | Retrieve document data using its identifier (name, space, form) |
create_document | Create a new document in a space, including its form and fields |
update_document | Update the metadata of an existing document |
search_knowledge_base | Ask natural-language questions about a knowledge base and receive the most relevant documents |
One of the most powerful aspects of Athento’s MCP implementation is that it has context about the configuration that exists within the document management system.
You do not need to explain that you have employee records or vehicle records, because the agent will discover that by itself.
When the client or agent requests the list of tools from Athento, Athento provides the teams that the authenticated user has permission to access. For each team, Athento’s MCP returns the configuration schema: which spaces, document types, and knowledge bases are available to that team. This way, without anyone having to teach it, the agent can speak the same language as the user interacting with it through Claude, ChatGPT, or any other technology.
In addition:
For now, the scope includes search, reading, document creation, metadata updates, and knowledge base queries. Document deletion and workflow management through MCP are not yet available, although these capabilities will be incorporated in the coming weeks.
This is where things get interesting.
With MCP enabled, your document management system stops being a destination that users need to visit to find information and becomes a resource that your AI agents can use autonomously within any conversation or workflow.
Some concrete examples:
Beyond these specific use cases, what really matters is the shift in the operating model: any business process that currently depends on someone opening the document management system to search for, consult, or enter data is a candidate for an AI agent to perform that task on their behalf, with the same permissions and traceability as always.
You can now connect your Athento instance—your document management system with MCP—to both Claude and ChatGPT in just a couple of minutes and without technical knowledge. In both cases, the pattern is the same: an OAuth application is created in Athento, the connector is configured in the AI client using the URL of your MCP server, and each user authorizes their own connection.
An organization administrator in Claude accesses the connector settings, adds a new “web” connector, and configures the MCP server URL for your instance—something like https://yourcompany.athento.com/mcp—together with the Client ID and Client Secret of an OAuth application previously created in Athento. Once the connector has been created, each user activates it from their own settings in Claude Desktop and authorizes access through the browser. From that point on, the connection is ready.
Complete guide: How to Connect Claude AI to Athento’s MCP
The process is similar.
It is also configured through Django OAuth Toolkit: a confidential OAuth application is created in Athento’s administration interface, the callback generated by ChatGPT when creating the connector is registered, and the authorization and token URLs (/o/authorize/ and /o/token/) are configured together with the MCP URL (/mcp) in the application settings in ChatGPT. When connecting, ChatGPT redirects the user to the Athento login page. The user authorizes access and can then ask ChatGPT things such as “find the latest invoice created” or “update the status of this invoice to paid.”
Complete guide: How to Use the Athento App in ChatGPT
This is only the beginning.
Athento’s MCP was launched with a deliberately limited scope—searching, reading, creating, and updating documents, plus querying knowledge bases—because we prefer to launch something solid and secure as early as possible and expand it based on real use cases requested by customers as they use it.
It is the same philosophy we use to build the rest of Athento: minimum viable product, rapid iteration, and constant feedback.
And we already have the next steps prioritized in our product backlog. These are the features we are working on to expand what an AI agent can do with Athento through MCP:
query_atql): a new tool that translates natural-language requests into ATQL queries, allowing users to ask agents to perform much more complex searches.set_lifecycle_state): the first workflow management capability within MCP, allowing an agent to change the state of a document within its lifecycle.get_document: an enhancement that will allow the agent to receive all document metadata when retrieving a document, rather than only basic information such as name, series, and document type.With these capabilities, Athento’s MCP will begin to cover not only document search and management but also some of the workflows that currently live within the platform. We will therefore be updating this article very soon with the latest developments.
If your team is already experimenting with AI agents and wants those agents to work with real document information, while maintaining permissions and traceability, now is the time to try it.
Do you already use Athento and want to enable your document management system with MCP in your instance? Contact us and we’ll help you configure it.
🤖Learn how to automate document management with ChatGPT by integrating it with Athento. Save time, reduce errors and improve your processes 🚀#Automation #ArtificialIntelligence.
Discover how you can chat with a PDF and get instant responses with AI. With Athento, you can chat with your documents and generate new ones automatically – transform your document management! 🤖🚀
Déjanos tus datos y te enviaremos el acceso por email
De conformidad con el art. 13 del RGPD y el art. 11 de la LOPDGDD, tus datos personales van a ser tratados por ATHENTO EUROPA S.L. con la finalidad de atender las consultas realizadas por los usuarios, siendo el consentimiento nuestra base legal para el tratamiento de tus datos personales. Sólo trataremos los datos que nos proporciones y podrán ser cedidos a terceros para poder cumplir la finalidad, como a ActiveCampaig u Open AI. Existen automatizaciones a través de Open AI, no obstante, hay intervención por lo que no se trata de decisiones automatizadas. Puedes ejercer tus derechos de acceso, rectificación, oposición, supresión, portabilidad y limitación de sus datos de carácter personal a través del email: dpo@athento.com. Puedes consultar la información adicional y detallada sobre Protección de Datos en nuestra Política de Privacidad.