Workflows#

This walkthrough chains together the steps from Getting started, Building flows, Components, and Deploying flows into one end-to-end scenario: building a chat flow that uses DKubeX models, testing it, and deploying it as an API endpoint.

Build and deploy a chat flow with DKubeX models#

  1. Launch Langflow. Click the Langflow tile in your DKubeX dashboard — you are logged in automatically via DKubeX SSO.

  2. Create a new flow. Click New Flow from the projects list and choose a blank canvas.

  3. Add the components. From the left sidebar, drag a Chat Input and a Chat Output (both in the Input & Output category) onto the canvas, then open the DKubeX Providers category and drag a DKubeX LLM component alongside them.

  4. Configure the DKubeX LLM. Click the component to open its side panel, enter your SecureLLM API Key, click Model Name and then Refresh list to load the available models, and select a model.

  5. Connect the components. Drag from each component’s output port to a compatible input port (compatible ports share a colour) to wire Chat Input → DKubeX LLM → Chat Output.

  6. Test the flow. Click the Playground button in the top-left corner of the canvas to chat with your flow and see its output. You can also click Run Component on any component to execute the flow up to that point.

  7. Deploy the flow. In Development Mode, open the three-dots ⋯ menu next to the flow name and click Deploy. In the Deploy Workflow modal, enter a Name (4–30 characters) and an optional Description, then click Deploy. You are switched to Deployment Mode, where the new deployment appears in the table.

  8. Call the endpoint. Once the deployment’s status is Deployed, click its row and open the API Access tab for a ready-to-use call to your flow.

Tip: To reuse a secret such as your SecureLLM API Key across flows, store it once under Settings → Global Variables and reference it from the component. If you deploy a flow that uses global variables, make sure those variables are defined before deploying — see Building flows.

Extending to retrieval: To ground a flow in your own documents, add a DKubeX Embeddings component and connect it to a Vector Store component to index and retrieve documents — see Components.