Cohere integration

What it does

The Cohere integration lets your agency invoke Command chat models, create embedding vectors with the Embed family, and rerank candidate documents against a query using the Rerank family, all from inside any workflow. Connect a client's Cohere key once and your workflows can summarize incoming form submissions, build a semantic index over a client's knowledge base, score retrieval candidates before handing them to another model, or generate structured JSON for downstream actions.

Connect a Cohere account

1. Open your workspace in TaskJuice and navigate to Connections.
2. Choose Cohere and click Connect.
3. In a new tab, open the Cohere API keys dashboard signed in as the client (or as your agency, if the client has delegated key creation to you).
4. Click New Trial Key or New Production Key, give it a name that identifies the workspace, and copy the key value.
5. Paste the key into TaskJuice and save the connection.

To rotate or revoke the key later, return to the Cohere API keys dashboard, delete the entry, generate a new key, and update the TaskJuice connection.

Triggers

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

Actions

• cohere/create-chat-completion generates a chat completion against a chosen Command model (for example, command-r-plus or command-r) with a list of messages and optional temperature, max tokens, top-p, top-k, stop sequences, and response-format controls.
• cohere/create-embedding returns one embedding per input string using a chosen Embed model such as embed-english-v3.0 or embed-multilingual-v3.0, with a required input_type hint and optional encoding selection.
• cohere/rerank-documents reorders a list of documents by relevance to a query using a Rerank model such as rerank-v3.5, optionally trimmed to the top N results.

Known limitations

• The integration calls the Cohere REST API with the connected key. Per-key rate limits, monthly quotas, and model-access entitlements are governed by the client's Cohere organization, not by TaskJuice. When Cohere returns a 429, TaskJuice surfaces it as a retryable rate-limit error and honors the recommended backoff.
• Streaming responses are not exposed. Each chat completion runs as a single request and resolves only after the full response is returned by Cohere.
• Embedding requests accept up to 96 texts per call. Larger batches must be split, typically by fanning out across a Loop node so each batch runs as its own request.
• The Embed v3 models require an input_type hint (search_document, search_query, classification, clustering, or image). Choose the value that matches how the embedding will be used downstream.
• 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 Cohere inside the workflow.