Langdock Integration

Yes. The connection is established via ContractHero’s MCP Connector—that is, via the Model Context Protocol. Langdock can create its own MCP connections, and ContractHero provides its contract data using this exact standard. In Langdock, enter ContractHero’s MCP URL, select OAuth as the authentication method, and log in once using your ContractHero credentials. After that, Langdock will answer contract-related questions directly in the chat, using data from your current contract portfolio. This requires that the MCP Connector be enabled in your ContractHero account.

You'll need three things: a ContractHero account with the MCP Connector enabled, the MCP URL, and a Langdock account that's authorized to create new MCP connections. The process is quick. In ContractHero, set up the connector on the Integrations page in the settings; in Langdock, create a new MCP connection, enter the URL, and select OAuth. After logging in, Langdock verifies the connection and displays the available queries. You don’t need an IT project, an in-house development, or a middleware system to do this.

No. The connector is technically limited to read-only access. You can retrieve data, but you cannot modify, delete, or create new data. This applies regardless of how a question is phrased in the chat: There are no write functions available through this method that Langdock could call. ContractHero therefore remains the primary source for your contracts, and there is no secondary data storage that could become out of sync.

No. The connector works with your account and, therefore, within the scope of your permissions. Roles and access rights from ContractHero remain unchanged. Contracts, categories, or fields that you cannot open in ContractHero will not be displayed in Langdock either. This means that the person responsible for procurement will not see employee contracts just because the question is asked in the chat. If you adjust permissions in ContractHero, this change will immediately affect the responses in Langdock as well.

Yes. The Model Context Protocol is an open standard, so you are not tied to a single AI provider. In addition to Langdock, the connector can be linked to Claude or ChatGPT, for example, as well as other assistants that support MCP. The process is always the same: enter the MCP URL, select an authentication method, and log in. You can also connect multiple assistants at the same time without changing any settings in ContractHero.

The connector starts with a deliberately small set of queries, such as retrieving a list of contracts. This allows questions like “Which contracts expire in the fourth quarter?” or “What contracts do we have with this supplier?” to be answered in the chat. The set is continuously expanded, and new queries become available automatically without requiring you to make any changes to the connection. During setup, Langdock shows which queries the connection currently provides.

No, it complements them. APIs and webhooks remain the go-to method when systems need to interact automatically—for example, for a nightly sync or a notification when a status changes. The MCP Connector is designed for the other scenario: a person has a question and asks it right where they’re already working—in this case, in Langdock. Both approaches can be used in parallel; they are not mutually exclusive.