Request BYO LLM setup and view your LLM configuration
Learn how to request Bring Your Own LLM (BYO LLM) setup for your AgentFlow tenant and view the LLM configurations in Admin Settings once the setup is complete.
Submit a support ticket with Reltio to initiate BYO LLM setup. To avoid delays, include all required details when you create the ticket.
Endpoint requirements
Your LLM endpoint must meet the following technical requirements before it can be integrated with AgentFlow:
| Requirement | Details |
|---|---|
| OpenAI-compatible API | Your endpoint must expose an OpenAI-compatible chat completions API (/v1/chat/completions or equivalent). AgentFlow sends requests in the OpenAI chat completions format and expects responses in the same format. |
| TLS / HTTPS | All endpoints must be served over HTTPS with valid TLS certificates. Production environments require certificates issued by a trusted certificate authority. |
| Authentication | Supported authentication types: OAuth2 client credentials, API key, or AWS access key/secret key (AKSK). If your endpoint uses OAuth2, you must provide a token URL and client credentials. |
| Network accessibility | Your endpoint must be reachable from Reltio's cloud infrastructure over the public internet using HTTPS. If your endpoint requires IP allowlisting, Reltio will provide the egress IP addresses to add to your allowlist. |
| Streaming | Your endpoint must support streaming responses (server-sent events). AgentFlow uses streaming to deliver model responses to users in real time. |
Prerequisites
Before submitting your support ticket, gather the following information for each environment (DEV, TEST, PROD) where you want BYO LLM enabled:
| Information | Description | Example |
|---|---|---|
| Tenant ID | Your Reltio tenant identifier | xxxxxxxx-prod |
| Environment(s) | Which environment(s) to configure | DEV, TEST, PROD |
| LLM provider name | The name of your LLM provider or internal gateway | "Acme Corp AI Gateway" |
| Chat completions endpoint URL | The OpenAI-compatible chat completions endpoint URL (per environment) | https://ai-gateway.acme.com/v1/chat/completions |
| Authentication endpoint URL | The token URL for OAuth2 authentication, if applicable (per environment) | https://auth.acme.com/oauth2/token |
| Authentication type | How your endpoint authenticates requests | OAuth2 client credentials, API key, AKSK |
| Model identifier(s) | The model name(s) available at your endpoint | gpt-4o, claude-3.5-sonnet |
| Streaming support | Whether your endpoint supports streaming responses, per model | Yes / No |
| IP allowlisting required? | Whether your endpoint requires Reltio's egress IPs to be allowlisted | Yes / No |
Step 1: Submit a support ticket
Open a ticket through the Reltio Support portal.
In your ticket, add the following details:
- Subject: BYO LLM Configuration Request — [Your Organization Name]
- Body: Include all the prerequisite information listed above. Provide details per environment if they differ across DEV, TEST, and PROD.
- Do not include credentials in the ticket. State that credentials will be shared securely after the ticket is assigned.
Step 2: Share credentials securely
After your ticket is assigned to a support engineer, share the following credentials using secure or encrypted email directly with the assigned engineer:
- Client ID
- Client secret (or API key)
- Any additional authentication tokens or certificates required by your endpoint
Step 3: Reltio configures and validates the integration
The Reltio team will:
- Configure the LLM endpoint(s) for your tenant in each requested environment.
- Validate connectivity and authentication against your endpoint.
- Run test calls to confirm the endpoint is reachable and returning responses correctly.
- Verify that no data is routed to Reltio-managed LLM instances for your tenant.
If your endpoint requires IP allowlisting, the Reltio team will provide the egress IP addresses that need to be added to your allowlist.
If any issues arise during validation (for example: authentication failures, network connectivity, or endpoint response errors), the Reltio team will work with you to resolve them.
Step 4: Activation
Once validation is complete, the Reltio team enables BYO LLM for your tenant. You will be notified when the configuration is active and ready for use. At this point:
- Your System Administrator can view the LLM configuration in AgentFlow Admin Settings.
- Data Stewards and Data Product Owners can select from the configured models when starting new conversations. For more information, see Select a model in a conversation.
View LLM configurations (System Administrator)
After BYO LLM is enabled, System Administrators can view the current LLM configurations in AgentFlow.
- Open AgentFlow and navigate to .
- Locate the Large Language Model Config section.
An information banner displays at the top of the section:
LLM configurations for this tenant are managed by Reltio. If you need changes or additional models, please contact the support team.
The configuration table displays the following for each LLM:
| Column | Description |
|---|---|
| Configuration name | The display name for the LLM configuration (for example: "Acme Corp AI Gateway") |
| Provider | The LLM provider type (for example: openai, bedrock, anthropic) |
| Available models | The model identifier(s) available at this endpoint (for example: gpt-4o) |
| Status | Active or Inactive, with a visual indicator |
If no LLM configurations exist for your tenant, this section displays: No LLM configurations available, please contact support.
Request configuration changes
All changes to BYO LLM configurations are handled by the Reltio team. To request a change, submit a support ticket through Reltio Support with the following details:
| What to include | Description |
|---|---|
| Tenant ID | Your Reltio tenant identifier |
| Environment | Which environment (DEV, TEST, PROD) the change applies to |
| Type of change | What you are requesting (see examples below) |
| Details | Specific information about the change |
Examples of changes you can request:
| Change type | What to include in the ticket |
|---|---|
| Add a new model | Model identifier, endpoint URL (if different from existing), authentication details. Share new credentials using secure email after ticket assignment. |
| Rotate or update credentials | State which configuration needs credential rotation. Share new credentials using secure email after ticket assignment — do not include credentials in the ticket. |
| Enable or disable a configuration | Configuration name and whether you want it enabled or disabled |
| Change endpoint URLs | New endpoint URL(s), which environment, and reason for change (for example: environment migration) |
| Remove a configuration | Configuration name and environment. Note: active conversations using this configuration will show an error. |
Permissions
| Role | What they can do |
|---|---|
System Administrator (ROLE_TENANT_ADMIN) | View LLM configurations and their status in Admin Settings (read-only). Cannot create, edit, enable, disable, or delete configurations. |
| AgentFlow user (any authenticated user) | Select a model in the chat input area when starting a new conversation. |
| Reltio delivery / engineering (internal) | Create, edit, enable, disable, and delete LLM configurations via backend tooling. |
Technical details
- Setup type
- BYO LLM is a one-time setup per environment (DEV, TEST, PROD). Once configured, the integration is active until explicitly changed. Credential rotation is periodic and follows the same support ticket flow.
- Billing
- BYO LLM has separate, lower-cost pricing. Your organization bears its own LLM provider costs (inference, tokens, API usage). Reltio charges only for platform usage (compute, networking, observability). For more information, refer to your AgentFlow pricing agreement.