Overview
Bud AI Foundry Router lets teams build reusable routing graphs for AI traffic. A router combines signals, projections, decision rules, algorithms, and plugins so a single OpenAI-compatible request can be evaluated and sent to the best endpoint or policy path. Routers are created as project resources, configured in a visual editor, saved as active router definitions, and exposed through the chat completions API by using the router name as themodel value.
Key Features
Visual Router Editor Build router graphs with drag-and-drop actions, category-aware lanes, and guarded connection rules. Project-Scoped Creation Create routers globally and assign them to a project, or create routers directly from project context. Dynamic Action Catalog Router actions load from the/routers/actions API and include parameter metadata, output definitions, validation rules, and categories.
OpenAI-Compatible Usage
Use a router through v1/chat/completions with your API key and the router name in the model field.
Operational Controls
Search, paginate, edit, delete, save, and inspect router cards with status, step count, run count, and last execution metadata.
Router Building Blocks
Router action categories are resolved from API metadata and action type prefixes. The editor blocks invalid edges and shows contextual guidance when users try to connect unsupported node pairs.
How It Works
- Create - Define the router name, description, icon, and project.
- Design - Open the router detail page and add actions from the router action catalog.
- Connect - Wire Trigger → Signal / Decision / Projection and Decision → Algorithm / Plugin.
- Configure - Fill action parameters using clusters, models, projects, providers, credentials, and endpoints.
- Save - Persist the router DAG as
dag_configand activate the router. - Use - Call
v1/chat/completionswithmodelset to the router name.