Developer Tools & DevOps

Connect Replicate to Your AI Agents

Let your Wkil agents run AI model predictions, track their status, and launch fine-tuning jobs through Replicate.

Authentication: API key

What you can do

Available capabilities may vary by authentication and workspace configuration.

Create Prediction

Run a model to generate a prediction from given inputs.

Get Prediction

Check the status and output of a running prediction.

Cancel Prediction

Stop a prediction that is still running.

List Public Models

Browse publicly available models on Replicate.

Create Training Job

Start a fine-tuning job for a specific model version.

An agent generates an image from a text prompt using a public Replicate model.
An agent checks whether a long-running prediction has finished before notifying a user.

1
Choose the integration
Add Replicate to your agent from inside the Wkil platform.
2
Grant the permissions it needs
Define precisely what the agent can access and perform.
3
Test, then run it
Test actions before they go live, with human approval where needed.

Permissions always stay in your control — you can scope access and require human approval before any sensitive action.

Does this use my own Replicate account?

Yes — the connection authenticates with an API key scoped to your own account.