Generating workflows using AI
You can use IBM® watsonx.ai to generate workflows in Concert Workflows by describing the task you want to automate. The workflow is then automatically generated and you can save it and customize it in the workflow editor.
Attention: This feature is available as a preview only.
Before you begin
- You must install the Concert Workflows add-on or a standalone version of Concert Workflows. These features are available for on-premises Concert deployments only.
- You must have installed IBM watsonx.ai and connected to your Concert instance. Refer to Implementing IBM watsonx.ai (on-premises) for details.
Note: The following instructions apply if you are using the embedded version of Concert Workflows as opposed to a standalone instance. If you
are using the standalone version, refer to Creating workflows in the IBM Rapid Infrastructure
Automation documentation.
Step 1: Enable IBM watsonx.ai for Concert Workflows
You must enable your Concert Workflows add-on to use IBM watsonx.ai. Refer to the following instructions:
- On the host where Concert Workflows is deployed,
navigate to the folder containing the Concert Workflows installation script. For OCP-based installations, this is the
deploy-k8s
script. For VM-based deployments, this is thesetup
script. - Create a
.yaml
file to specify the custom values.vi my-custom-values.yaml
- In the new file, add a section called
rna.instance.ai
, and then add the AI-specific attributes and values.Table 1. Required attributes to enable IBM watsonx.ai to generate workflows. The following table describes the required attributes to enable AI for Concert Workflows. Attribute Description watsonx_version
The version of IBM watsonx.ai to use with Concert Workflows. Set this attribute value to "2024-03-14"
.watsonx_model
The IBM watsonx.ai model to use to generate workflows. Set this value to "meta-llama/llama-3-3-70b-instruct"
.watsonx_auth_type
The authentication method used by Concert Workflows to authenticate IBM watsonx.ai. - For Identity and Access Management (IAM) authentication, set this value to
"iam"
. - For authentication via IBM Cloud Pak® for Data, set this value to
"cp4d"
. - For bearer token authentication, set this value to
"bearertoken"
. - To disable this feature, set this value to
"disabled"
.
Attention: To enable Concert Workflows to use IBM watsonx.ai to generate workflows, you must set this attribute to"iam"
,"bearertoken"
, or"cp4d"
. Based on your selection, you must set additional attributes. Refer to Table 2.watsonx_cluster_url
The URL of the IBM watsonx.ai cluster. watsonx_project_id
,watsonx_space_id
, orwatsonx_deployment_id
Specify one of the following ID values for the IBM watsonx.ai project, deployment space, or deployment that you want to use. Table 2. Additional authentication attributes Authentication method Required attributes Description If you are authenticating using Identity and Access Management ( "iam"
)...watsonx_api_key
IBM Cloud API key that is used to authenticate to IBM watsonx.ai If you are authenticating using IBM Cloud Pak for Data ( "cp4d"
)...watsonx_username
Username that is used by IBM Cloud Pak for Data to authenticate to IBM watsonx.ai. watsonx_password
Password that is used by IBM Cloud Pak for Data to authenticate to IBM watsonx.ai. cp4d_cluster_url
URL of the IBM Cloud Pak for Data instance. If you are authenticating using a bearer token ( "bearertoken"
)...watsonx_bearer_token
Bearer token value - For Identity and Access Management (IAM) authentication, set this value to
- Upgrade your Concert Workflows add-on to enable
it to use IBM
watsonx.ai.
- If deployed on an OCP cluster, run the following command, replacing
<custom-YAML-file>
with the name of the custom attribute file that you just created../deploy-k8s --override-values=<custom-YAML-file>
- If deployed on a virtual machine, run the following command, replacing
<custom-YAML-file>
with the name of the custom attribute file that you just created./setup --override-values=<custom-YAML-file>
- If deployed on an OCP cluster, run the following command, replacing
Step 2: Generate a workflow using AI
- Click from the main navigation.
- Click Create workflow.
- Click Generate with AI.
- In the prompt input field, describe the task you want to automate. For example, you can test the
functionality with a basic request, such as: Reverse the elements in this array: [9, 8, 7, 6,
5].
The workflow generation process begins. In the preview area, status messages are displayed at each stage of the generation process. When the process is complete, the generated workflow is shown.
- Click Save and open in editor to modify the workflow. Refer to Using the workflow editor for details.
Each prompt provided generates a new workflow. When generated, you can update your workflow in
the workflow editor.
Note: You cannot modify workflows directly in the Generate workflow
with AI page.
To switch between multiple generated workflows, click the Load automation preview icons in the chat history.