← All Plugins
Integration

Backstage <> AI EG

Self-service AI access from your developer portal.

Backstage <> AI EG brings Agent Router (EAIG) resources into Backstage. Developers discover approved models and AI APIs in the service catalog, request access, and see routing and policy, while platform teams keep control of the gateway.

Talk to an Agent Router (EAIG) ExpertSee How It Works

What Backstage <> AI EG does

What it does

The integration reads AI gateway resources from Agent Router, such as routes, AI service backends, models, and attached policies, and publishes them as Backstage catalog entities. A Backstage plugin adds AI-specific views and self-service actions, and scaffolder templates create the gateway configuration when a developer onboards.

The enterprise problem

Developers who want to use AI often do not know which models are approved, which endpoint to call, or how to get credentials, so they open tickets or go straight to a provider with a personal key. Platform teams lose visibility and governance, and onboarding takes days.

Where it fits in Agent Router

Backstage becomes the front door and Agent Router remains the enforcement layer. Backstage shows and requests AI gateway resources; Agent Router applies the routing, security, and quota policies behind them. The gateway configuration stays the source of truth.

How It Works

The request and response flow

  1. 1
    Control

    Sync gateway resources to the catalog

    An entity provider reads Agent Router routes, backends, models, and policies and registers them as Backstage API and resource entities.

  2. 2
    Request

    Developer discovers AI APIs

    In Backstage, a developer browses approved models and AI APIs, with owners, documentation, pricing tier, and attached policies.

  3. 3
    Request

    Self-service onboarding

    The developer runs a scaffolder template to request access for their component, choosing the model, environment, and quota tier.

  4. 4
    Control

    Gateway configuration is generated

    The template opens a pull request (or applies directly) with the AI route, credentials, and quota policy for Agent Router.

  5. 5
    Response

    Visibility in the portal

    Once live, the component page shows its AI endpoint, routing, policies, and usage from Agent Router.

Architecture

Backstage <> AI EG in Agent Router

Backstage <> AI EG architecture: App developers, Data scientists, Platform engineers, Service owners send requests to Agent Router, which applies Backstage catalog & plugin, then AI gateway resources, then Routes · policies · quotas, and connects to OpenAI, Anthropic, Amazon Bedrock, Self-hosted vLLM.DeveloperBackstage developer portalService catalogAI APIs · owners · docsModel catalogApproved models · cost tierScaffolder templatesSelf-service AI accessCatalog syncGitOps pull requestAI gateway resourcesAIGatewayRouteRoutes & modelsAIServiceBackendProvider backendsPoliciesAuthN/Z · guardrails · quotasApplied to the data planeAgent RouterEnforces routing, security, and quotas on every requestOpenAIAnthropicAmazon BedrockSelf-hosted vLLM
Developers
App developersData scientistsPlatform engineersService owners
Agent Router
Backstage <> AI EG plugin
Backstage catalog & plugin
AI gateway resources
Routes · policies · quotas
Models & providers
OpenAIApproved model
AnthropicApproved model
Amazon BedrockApproved model
Self-hosted vLLMPrivate model

Developers work in Backstage. Agent Router resources back the catalog and enforce every request to models and providers.

Key Features

Built for production AI traffic

AI API discovery

Publish Agent Router routes as Backstage API entities with OpenAPI specs, owners, and documentation.

Model catalog

List approved models and providers with capabilities, regions, and cost tier so developers choose the right one.

Self-service onboarding

Scaffolder templates create AI routes, credentials, and quotas through GitOps pull requests.

Routing visibility

See which providers and fallbacks serve each AI API directly on its catalog page.

Policy management

View attached authentication, guardrail, and quota policies, and request changes from the portal.

Ownership and usage

Link AI consumers to owning teams and show their usage and budget from Agent Router.

Benefits

Engineering and business outcomes

Developer experience

Onboarding in minutes

Developers find and request AI access in the portal they already use, without tickets or waiting.

Governance

Approved paths by default

The easiest way to use AI is also the governed way, which reduces shadow AI usage with personal keys.

Operational efficiency

Less platform toil

Templates and GitOps replace manual gateway configuration for every new consumer.

Visibility

One map of AI usage

Platform leads can see which services use which models, owned by which teams.

Use Cases

Where enterprises use Backstage <> AI EG

01

Internal AI platform launch

Expose a curated set of models to hundreds of engineering teams through a single portal experience.

02

Golden-path AI onboarding

A template gives every new service an AI route with standard authentication, guardrails, and quota.

03

AI inventory for security reviews

Security teams use the catalog to see every component that calls an AI model and which policies apply.

04

Model deprecation

Find every consumer of a model being retired from its catalog page and notify the owning teams.

Configuration

AI API entity and onboarding template

Catalog syncWhich gateway resources appear in the Backstage catalog.
All Agent Router routes in ai-platform
Entity typeHow AI APIs are represented, with specs, owners, and docs.
Backstage API (OpenAPI)
Linked policiesPolicies shown on each AI API's catalog page.
AuthN/Z · Guardrails · Token quota
Onboarding templateSelf-service form developers use to get access.
Request AI access for a component
Template choicesWhat developers pick when requesting access.
Model, environment, quota tier
DeliveryHow the generated Agent Router configuration is applied and reviewed.
GitOps pull request
OwnerTeam responsible for approving access and maintaining the AI APIs.
group:ai-platform

Illustrative example. Settings and values show what can be configured and are not exact configuration syntax.

Enterprise Workflow

How it fits your AI platform

  1. 01

    Developers

    Discover AI APIs and models in Backstage and request access with a template.

  2. 02

    Backstage

    Shows the AI catalog, runs onboarding templates, and displays routing, policy, and usage.

  3. 03

    Platform team

    Curates approved models, owns templates, and reviews generated gateway changes.

  4. 04

    Agent Router

    Holds the AI gateway resources and enforces routing, security, and quota on every request.

  5. 05

    Models & providers

    Serve traffic only through the governed routes created from the portal.

Related Plugins
AI Optimization

Semantic Caching

Answer repeat questions from cache, not from the model.

View plugin →
AI Governance

Quota Management & Tokenomics

Give every team a token budget, and enforce it at the gateway.

View plugin →
AI Security

Security Guardrails

Inspect every prompt and every response before it causes harm.

View plugin →
AI Security

AuthN/Z

Know who is calling your models, and control what they can use.

View plugin →
AI Routing

Info-Aware Routing

Send every request to the best provider right now.

View plugin →
Browse All Plugins

Make AI self-service for every developer

IMESH can help you connect Backstage to Agent Router (EAIG), design golden-path templates, and build your internal AI platform.

Get Started with Agent Router (EAIG)Explore All Plugins