Eden AI integration

What it does

The Eden AI integration lets your agency call dozens of upstream AI providers (OpenAI, Anthropic, Google, Microsoft, Mistral, Cohere, AWS, DeepL, and more) through a single normalized API and a single key. Inside any workflow you can generate chat completions, embed text for semantic search, translate copy into the client's target market language, and pull text out of scanned invoices or contracts. Specify a primary provider per call, declare fallbacks for resilience, or fan out to several providers and let your workflow pick a winner.

Connect an Eden AI account

1. Open your workspace in TaskJuice and navigate to Connections.
2. Choose Eden AI and click Connect.
3. In a new tab, open the Eden AI dashboard signed in as the client (or as your agency, if the client has delegated key creation to you).
4. Go to Settings, API Keys, and copy the active key.
5. Paste the key into TaskJuice and save the connection.

To rotate the key later, generate a new one on the same page and update the TaskJuice connection. The old key keeps working until you revoke it.

Triggers

Eden AI does not publish a public webhook surface, so the integration is action-only. Use a Schedule or another app's trigger to invoke Eden AI actions inside a workflow.

Actions

• eden-ai/chat generates a chat completion against one or more upstream LLM providers with optional system prompt, prior history, temperature, max tokens, and a fallback provider list.
• eden-ai/embeddings returns embedding vectors for one or more input strings across one or more embedding providers, useful for semantic search and clustering.
• eden-ai/translate translates text between languages using Google, DeepL, Microsoft, or any other Eden AI translation provider, with optional source-language auto-detection.
• eden-ai/ocr runs optical character recognition on a hosted image or PDF URL and returns extracted text plus bounding boxes.

Known limitations

• Every Eden AI request is billed against the connected key's Eden AI account, and Eden AI in turn bills the upstream provider. Per-provider rate limits, monthly spend caps, and model entitlements are governed by the client's Eden AI organization, not by TaskJuice. When Eden AI returns a 429, TaskJuice surfaces it as a retryable rate-limit error.
• Streaming responses are not exposed. Each chat completion runs as a single request and resolves only after Eden AI returns the full response from the chosen upstream provider.
• OCR actions require a publicly accessible file URL. To run OCR on a private file, upload it to a signed-URL store first and pass the temporary URL.
• The integration is action-only. To react to events in real time, use a Schedule trigger, a polling trigger from another app, or an inbound webhook from a different source and call Eden AI inside the workflow.