Documentation

WorkBuddy

WorkBuddy can call custom models through an OpenAI-compatible provider. Add LLMPool as that provider and use a Product ID from the Board model catalog.

Configure a custom model

Before starting, create a Board API key, copy the OpenAI-compatible Base URL (for example https://your-board.example/openai/v1), and copy a model ID from Models.

In WorkBuddy, open the settings area named Models, Model providers, or Custom model (the label varies by version), then enter:

FieldValue
ProviderLLMPool
ProtocolOpenAI Compatible
Base URLhttps://api.benpay.ai/openai/v1; do not append /chat/completions
API keyYour Board API key
Model IDThe Product ID from Models
Context windowNo larger than the value shown for the model

If the form separates host and path, use /openai/v1 as the API path. If it asks for a complete request URL, use https://api.benpay.ai/openai/v1/chat/completions exactly once. Save, restart WorkBuddy if needed, and select the custom model.

Troubleshooting

  • 401: use a Board API key, not an Admin JWT or Tenant Access Token.
  • 404 or model_not_found: remove duplicated URL paths and copy the Product ID again from Models.
  • A visible model can still have no available capacity; check that its Product is active and has an available upstream.