Langflow Innovation Release

Langflow is an open-source tool to build and deploy AI pipelines and MCP servers. It supports important AI functionality like agents and the Model Context Protocol (MCP), and it doesn't require the use of specific large language models (LLMs) or vector stores.

EDB components in Langflow

You can build an AI pipeline by using the components available in the EDB folder in Langflow. To use the EDB components, make sure the required prerequisites are met before using them in Langflow.

Prerequisites

Users need the required permissions in HM to create AI pipelines with Langflow. You can manage users and permissions in the HM console.

• EDB Embedding and EDB Model Server components: Requires the AI Model Manager role at organization level:

  • Profile > User management > Assign Organization Roles > AI Model Manager.

• EDB Knowledge Base and EDB Database components: Require the Project Viewer role:

  • Project > Users > Assign Project Roles > Project Viewer.

• Create an access key by creating a machine user or via the My Account option in your profile.

• Access Langflow via Launchpad in the HM console.

• Save the access key and other variables in Langflow as Global Variables.

EDB Knowledge Base component

The EDB Knowledge Base component allows your flow to perform semantic similarity search using existing AIDB pipelines in your EDB cluster.

Prerequisites

• Create your access key and make sure the Project Viewer role is assigned to the machine user you created:

  • Project > Users > Assign Project Roles > Project Viewer.

• Create a cluster from a project and request access to the cluster with your machine user.

• Create a Knowledge Base via aidb in the cluster.

CREATE EXTENSION aidb CASCADE;

CREATE TABLE public.test_source_table_gt_prj1
(
    id               INT GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY,
    content          TEXT NOT NULL,
    unrelated_column TEXT
);

INSERT INTO public.test_source_table_gt_prj1
VALUES (43941, 'Catwalk Women Brown Heels'),
       (55018, 'Lakme 3 in 1 Orchid  Aqua Shine Lip Color'),
       (19337, 'United Colors of Benetton Men Stripes Black Jacket');

SELECT aidb.create_table_knowledge_base(
    name => 'psr_261_knowledge_base',
    model_name => 'bert',  -- this is a pre-defined locally running model
    source_table => 'public.test_source_table_gt_prj1',
    source_data_column => 'content',
    source_data_format => 'Text'
);

SELECT aidb.bulk_embedding('psr_261_knowledge_base');

--- below for Testing
select count(*) from public.test_source_table_gt_prj1;

SELECT * FROM aidb.kbs; 

SELECT * FROM aidb.kbstat; 

SELECT * FROM aidb.retrieve_key('psr_261_knowledge_base', 'orchid');
SELECT * FROM aidb.retrieve_text('psr_261_knowledge_base', 'orchid');

EDB Knowledge Base in Langflow

1. In Langflow, select EDB in the left side pane.

2. Drag and drop the EDB Knowledge Base Component to the flow.

3. In the field HM Machine User Key, enter the access key you have created.

4. Under Hybrid Manager Project, select your project from the dropdown menu.

5. Under Hybrid Manager Database, select the cluster that contains the EDB Knowledge Base.

6. Under Database Connection Type, select the connection type from the dropdown menu.

7. Under User for the database, select the user you saved in the global variables from the dropdown menu or type the user name in the field directly.

8. Under Password for database, select the password you saved in the global variables from the dropdown menu or type a password in the field directly.

9. Under Database Name, select the database where you created the Knowledge Base.

10. Under Query Value, enter a keyword of your choice.

11. Select the Play icon on the upper-right corner of the component.

12. Select Result Table (icon) in the lower-right corner to verify the component output.

EDB Database component

The EDB Database component is the standard data connector for your flow.

Prerequisites

• Make sure the Project Viewer role is assigned to the machine user you created to access the cluster:

  • Project > Users > Assign Project Roles > Project Viewer.

• Create a cluster.

EDB Database in Langflow

1. In Langflow, select EDB in the left side pane and move the EDB Database component per drag and drop to Langflow.

2. In the field HM Machine User Key, enter the Access key you have created.

3. Under Hybrid Manager Project, select your project from the dropdown menu.

4. Under Hybrid Manager Database, select the cluster that contains the EDB Knowledge Base.

5. Under Database Connection Type, select the connection type from the dropdown menu.

6. Under User for the database, select the user you saved in the global variables from the dropdown menu or type the user name in the field directly.

7. Under Password for database, select the password you saved in the global variables from the dropdown menu or type a password in the field directly.

8. Under Database Name, select the database you created the Knowledge Base in.

9. Select the Play icon on the upper-right corner of the component.

10. Select Result Table (icon) in the lower-right corner to verify the component output.

EDB MCP Server component in Langflow

Langflow supports the MCP Server for the EDB Database component.

Prerequisites

• Make sure the AI Model Manager role is assigned to the machine user you created:

  • Profile > User management > Assign Organization Roles > AI Model Manager.

• Set up the EDB DB Component.

• In Langflow > Profile > Settings > Langflow API Key > Add New > Generate API Key.

• Toggle MCP Server in the lower-right hand of the EDB Database component.

Add the EDB MCP Server component to the flow

1. In Langflow, select the MCP icon from the left navigation pane.

2. Drag and drop your MCP Server to the flow.

3. Under Tool, select the action you want to run.

4. Select the Global Variable you have created.

5. Select the Response icon in the lower-right corner of MCP Tools.

6. Run the MCP server in Tool mode to improve the functionality of the actions.

   • If you enable Tool mode, you can attach tools to the AI agent for additional and targeted functionality.

   • If you add an AI agent, the AI agent can use tools.

   • If you add the options Chat Input and Chat Output to the agent and select Playground, you can ask the AI agent to use the tool.

EDB Embeddings component

The EDB Embeddings component is a model-related component used to turn AI data into embedding vectors that also interacts with models hosted on EDB infrastructure.

It requires the creation of model clusters.

Prerequisites

You have to create a model cluster to use the EDB Embeddings component. Make sure the AI Model Manager role is assigned to the machine user you created:

• Profile > User management > Assign Organization Roles > AI Model Manager.

Creating a model cluster

1. In the HM console, select Asset Library in the upper-left hand.

2. In the Asset Library, select Models.

3. Select Add New Model on the right hand side of your screen.

4. Specify the configuration for the new model cluster in the provided parameter fields.

5. In the Models section, select the model cluster you have created.

6. In the upper-right hand, select Create Cluster With This Model.

7. Enter a name for the model cluster.

8. In the Asset Library, select the model cluster you have created.

9. Under Clusters Using This Model, check Status to verify that the cluster is active.

EDB Embeddings component in Langflow

1. In Langflow, select EDB in the left side pane.

2. Drag and drop the EDB Embeddings Component to the flow.

3. In the field HM Machine User Key, enter the access key you have created.

4. Under Hybrid Manager Model Server Cluster Instance, select the model cluster you have created.

5. Select the Play icon in the upper-right corner of the component.

EDB Model Server component

The EDB Model Server component acts as the interface for interacting with AI models hosted on EDB infrastructure. Types of models include LLMs, embedding models, and other inference models.

The EDB Model Server component requires the creation of model clusters.

EDB Model Server component in Langflow

1. In Langflow, select EDB in the left side pane.

2. Drag and drop the EDB Model Server Component to the flow.

3. In the EDB Model Server component UI, enter an input in the field Input.

   For example: "Hi".

4. In the lower-right corner, make sure to select Model Response.

5. Select the Play icon in the upper-right corner of the component.

6. Select the Result icon in the lower-right corner to verify output.