--- title: "Connect Navan to ChatGPT: Manage Traveler Profiles and Accounts" slug: connect-navan-to-chatgpt-manage-traveler-profiles-and-accounts date: 2026-09-16 author: Nachi Raman categories: ["AI & Agents"] excerpt: "Learn how to connect Navan to ChatGPT using a managed MCP server. Execute traveler profile management, SCIM directory syncs, and account workflows using natural language." tldr: "This guide covers how to generate a secure MCP server for Navan using Truto, handle Navan's TMC entitlement quirks via SCIM, and connect the API directly to ChatGPT for autonomous travel management." canonical: https://truto.one/blog/connect-navan-to-chatgpt-manage-traveler-profiles-and-accounts/ --- # Connect Navan to ChatGPT: Manage Traveler Profiles and Accounts If you need to connect Navan to ChatGPT to automate corporate travel administration, orchestrate SCIM directory syncs, or manage traveler profiles, you need a [Model Context Protocol (MCP) server](https://truto.one/what-is-mcp-and-mcp-servers-and-how-do-they-work/). This server acts as the translation layer, exposing Navan's complex travel and expense schemas as discoverable JSON-RPC tools that ChatGPT can safely call. If your team uses Claude, check out our guide on [connecting Navan to Claude](https://truto.one/connect-navan-to-claude-lookup-user-directory-and-account-data/) or explore our broader architectural overview on [connecting Navan to AI Agents](https://truto.one/connect-navan-to-ai-agents-access-traveler-identity-and-profiles/). Giving a Large Language Model (LLM) read and write access to a global travel management platform is an engineering challenge. Navan enforces strict data models for cost centers, legal entities, and traveler entitlements. You either spend weeks building and hosting a custom integration service to map LLM arguments to Navan's payloads, or you use a [managed infrastructure layer](https://truto.one/best-mcp-server-platforms-for-enterprise-ai-agents-2026/) to dynamically generate a secure MCP server. This guide breaks down exactly how to use Truto to generate an authenticated MCP server for Navan, connect it natively to ChatGPT, and execute complex user management workflows using natural language. ::cta{buttonText="Talk to us" buttonUrl="/book-a-demo/"} Stop writing boilerplate API integration code. Let Truto generate secure, [managed MCP servers](https://truto.one/best-mcp-server-platforms-for-enterprise-ai-agents-2026/) for your AI agents in seconds. ::: ## The Engineering Reality of the Navan API Building a custom MCP server means you own the entire integration lifecycle. While the [open MCP standard](https://truto.one/what-is-mcp-model-context-protocol-the-2026-guide-for-saas-pms/) dictates how ChatGPT discovers tools, implementing those tools against Navan's highly specific API surface reveals several critical quirks. If you build this in-house, here are the Navan-specific engineering realities your server must handle: ### The TMC Entitlement Trap (403 Forbidden) Navan gates generic user directory data behind a Partner/TMC (Travel Management Company) entitlement. If you generate standard Navan API credentials from the `Travel > Settings > Integrations` dashboard and attempt to call the `/users` endpoints to list profiles, Navan will return a hard `403 Forbidden`. These standard credentials simply do not carry the entitlement required to read the global employee directory. To solve this and manage traveler profiles programmatically, you must bypass the standard travel API and instead connect via the **Navan SCIM integration** (`navanscim`). Your MCP server must be smart enough to route directory queries through the SCIM standard rather than the primary travel endpoints. ### Hard-Failing Validation on Schema Enums When an LLM attempts to update a user's department, cost center, or travel policy, Navan expects exact, case-sensitive string matches that correlate to your company's predefined lists. Navan's API does not perform fuzzy matching. If ChatGPT hallucinates a currency code or passes an invalid locale string, the API drops the request. Your MCP server must expose highly detailed JSON Schemas - directly derived from documentation - so the LLM knows exactly which enumerations are valid. ### Upstream Rate Limits and IETF Headers Navan enforces rate limits to protect its infrastructure. A common mistake developers make when building custom MCP servers is attempting to swallow these errors, artificially sleeping the server thread, and retrying. Truto takes a different, standards-based approach: **Truto does not retry, throttle, or apply backoff on rate limit errors.** When the Navan API returns an HTTP 429, Truto immediately passes that error back to the caller. However, Truto normalizes the upstream rate limit information into standard headers (`ratelimit-limit`, `ratelimit-remaining`, `ratelimit-reset`) per the IETF specification. It is the responsibility of the caller (the AI agent) to read these headers, wait until the reset window clears, and retry. Do not attempt to mask rate limits from the LLM; surface them so the agent can plan its execution schedule. ## How to Build and Connect the Navan MCP Server Truto completely eliminates the need to [manually code tool definitions](https://truto.one/auto-generated-mcp-tools-for-ai-agents-a-2026-architecture-guide/). Because Truto derives tools dynamically from the Navan integration's underlying resources and documentation, your MCP server is generated on the fly. Here is how to create the server and plug it into ChatGPT. ### Step 1: Create the MCP Server Every MCP server is scoped to a single integrated account. This means the server URL inherently knows which Navan tenant to authenticate against. You can create this server via the Truto UI or programmatically via the API. **Method A: Via the Truto UI** 1. In your Truto dashboard, navigate to the **Integrated Accounts** page for your Navan connection (remember to use the SCIM connection if you are focusing on user management). 2. Click on the **MCP Servers** tab. 3. Click **Create MCP Server**. 4. Select your configuration options (e.g., restricting to `read` methods or filtering by specific tags like `users`). 5. Click Save and immediately copy the generated MCP server URL. Treat this URL as a secret. **Method B: Via the API** For dynamic deployments, you can issue a POST request to Truto to provision the server. This is ideal if you are generating unique endpoints for different enterprise tenants. ```bash curl -X POST https://api.truto.one/integrated-account//mcp \ -H "Authorization: Bearer $TRUTO_API_TOKEN" \ -H "Content-Type: application/json" \ -d '{ "name": "Navan Directory for ChatGPT", "config": { "methods": ["read", "write"], "tags": ["users", "cost_centers"] } }' ``` The response contains the cryptographic token URL. This URL handles JSON-RPC translation, Navan authentication, and schema validation automatically. ### Step 2: Connect the Server to ChatGPT Once you have the Truto MCP URL, you must register it with ChatGPT. You can do this through the OpenAI interface or via a local configuration file if you are running a custom desktop agent. **Method A: Via the ChatGPT UI** 1. Open ChatGPT and navigate to **Settings -> Apps -> Advanced settings**. 2. Enable **Developer mode** (MCP capabilities require this flag to be active). 3. Under the Custom Connectors or MCP servers section, click to add a new server. 4. Enter a recognizable name (e.g., "Navan SCIM via Truto"). 5. Paste the Truto MCP URL into the Server URL field and click **Add**. ChatGPT will immediately ping the endpoint, execute an `initialize` handshake, and request the `tools/list`. The Navan tools will now be available in your chat context. **Method B: Via Manual Config File (for local dev/Desktop)** If you are running a local setup or a framework that requires standard MCP configuration files, you can use the official SSE transport package. Create or update your JSON configuration: ```json { "mcpServers": { "navan-truto": { "command": "npx", "args": [ "-y", "@modelcontextprotocol/server-sse", "https://api.truto.one/mcp/" ] } } } ``` ## Hero Tools for Navan When ChatGPT issues a `tools/list` request, Truto dynamically derives the tools from Navan's documentation schema. All query and body parameters share a flat input namespace, which Truto intelligently parses based on the integration's underlying definitions. Here are the highest-leverage operations for automating Navan traveler management. ### list_all_navan_users Retrieves a paginated list of all active profiles within the company's Navan instance. As mentioned, because Navan gates general user data behind TMC entitlements, standard API credentials will return a 403 error. You must connect the Navan SCIM integration to successfully execute this tool. > "Get a list of all active employees currently registered in the Navan travel directory. Return only their names, emails, and their assigned cost centers. If there is a pagination cursor, fetch the next page until you have all records." ### get_single_navan_user_by_id Fetches the complete profile payload for a specific user. This is critical for auditing an individual's travel policies, checking their assigned approval manager, or validating their primary legal entity before booking flights. > "Look up the Navan profile for the user ID 'usr_892nf82'. Tell me who their designated approval manager is and what travel policy they are assigned to." ### create_a_navan_user Provisions a new traveler profile. This tool accepts a complex body schema requiring strict strings for parameters like locale, currency, and department mapping. It is frequently used during employee onboarding flows. > "Create a new Navan traveler profile for John Doe (john.doe@company.com). Assign him to the Engineering department, set his primary locale to en-US, and map him to the US-East cost center." ### update_a_navan_user_by_id Modifies an existing traveler's profile. This tool uses a PATCH method under the hood, allowing the AI agent to update specific fields - such as promoting a user to an administrative role or reassigning their cost center - without overwriting their entire profile. > "Update the Navan profile for user ID 'usr_892nf82'. Change their department to 'Product Management' and update their cost center code to 'CC-405'." ### list_all_navan_departments Fetches the allowed enumeration of departments configured in the Navan tenant. Because Navan requires exact string matches when assigning users to departments, ChatGPT should always call this tool first to prevent hallucinating an invalid department name. > "Fetch the list of all valid departments in our Navan account. Find the exact string ID for 'Enterprise Sales' so we can use it to update Jane's profile." ### get_single_navan_department_by_id Retrieves details about a specific department, including any budget limits or specific travel routing rules assigned to the group at the administrative level. > "Pull the configuration details for the department ID 'dept_5592'. Tell me if there are any specific policy overrides active for this group." For the complete tool inventory and detailed JSON schema requirements, visit the [Navan integration page](https://truto.one/integrations/detail/navan). ## Workflows in Action Giving ChatGPT access to these tools allows you to replace rigid scripts with autonomous, self-healing workflows. Here is exactly how an AI agent orchestrates these operations. ### Scenario 1: Auditing Unassigned Users and Fixing Cost Centers IT teams frequently need to ensure that every active traveler is tied to a valid billing code. If a user's cost center is missing, their travel expenses cannot be reconciled. > "Scan our Navan directory for any users who do not currently have a cost center assigned. Once you find them, cross-reference their department. If they are in the 'Engineering' department, update their profile to use cost center 'CC-101'." **Execution Steps:** 1. **`list_all_navan_users`**: ChatGPT calls this tool, looping through the `next_cursor` pagination to retrieve the full directory. 2. **Analysis**: The model filters the returned JSON array in memory, identifying profiles where the `cost_center` field is null. 3. **`list_all_navan_departments`**: To ensure it doesn't fail validation, the model fetches the exact department strings to confirm the ID for 'Engineering'. 4. **`update_a_navan_user_by_id`**: For every user matching the criteria, ChatGPT executes an update operation, passing the user's ID in the query and `{"cost_center": "CC-101"}` in the body. ```mermaid sequenceDiagram participant User as Human participant ChatGPT as ChatGPT participant MCP as Truto MCP Server participant Upstream as Upstream API (Navan) User->>ChatGPT: "Fix Engineering cost centers" ChatGPT->>MCP: Call list_all_navan_users MCP->>Upstream: GET /scim/v2/Users Upstream-->>MCP: Array of users MCP-->>ChatGPT: JSON result ChatGPT->>MCP: Call update_a_navan_user_by_id (usr_123) MCP->>Upstream: PATCH /scim/v2/Users/usr_123 Upstream-->>MCP: Success MCP-->>ChatGPT: Updated profile ChatGPT-->>User: "Updated 4 users to CC-101" ``` ### Scenario 2: Offboarding a User and Suspending Travel When an employee leaves the company, their ability to book corporate travel must be revoked immediately. > "We are offboarding Sarah Jenkins. Find her Navan profile using her email (sarah.j@company.com), suspend her account, and confirm when it is done." **Execution Steps:** 1. **`list_all_navan_users`**: ChatGPT queries the directory. (Note: Because query and body parameters share a flat namespace, if the schema supports an email filter, the model will pass it here to save bandwidth). 2. **`update_a_navan_user_by_id`**: ChatGPT extracts Sarah's user ID from the response and executes the update tool, setting `active: false` in the payload. 3. **Confirmation**: ChatGPT reads the successful response and reports back to the user that travel privileges have been revoked. ## Security and Access Control Exposing an enterprise travel platform to an AI agent requires strict guardrails. Truto provides several mechanisms to lock down the MCP server at the time of creation: * **Method Filtering**: Using the `config.methods` array, you can restrict the server to only specific operations. Setting `methods: ["read"]` ensures the LLM can only query data (`get` and `list`), making it physically impossible for the agent to modify a user profile. * **Tag Filtering**: Using `config.tags`, you can scope the available tools down to specific integration resources. If you only want the agent to touch directory information, you might restrict tags to `["users", "departments"]`, keeping the agent entirely isolated from expense and policy tools. * **Require API Token Auth**: By default, possessing the Truto MCP URL grants access. For higher security environments, setting `require_api_token_auth: true` forces the MCP client to also pass a valid Truto API session token in the authorization header, adding a secondary identity check. * **Expiration (`expires_at`)**: You can assign an ISO datetime to the token. Once reached, the underlying Key-Value store automatically purges the credential and a Durable Object alarm cleans up the database record, ensuring temporary AI access cannot become a permanent backdoor. ## Managing the Future of Travel Operations Connecting Navan to ChatGPT via MCP transforms static directory management into an interactive, conversational workflow. Instead of hunting through dashboards to verify approval hierarchies or writing custom scripts to handle SCIM provisioning, your IT and Finance teams can simply instruct their agents to audit, route, and correct traveler profiles on the fly. By leveraging Truto's [dynamic tool generation](https://truto.one/auto-generated-mcp-tools-for-ai-agents-a-2026-architecture-guide/), you sidestep the TMC entitlement traps, avoid writing boilerplate JSON-RPC handlers, and ensure your LLMs receive the strict schemas they need to operate safely. The result is a resilient integration that adapts instantly to documentation updates and scales effortlessly alongside your enterprise workflows.