Build an AI Agent with MindsDB

MindsDB provides a custom syntax to build AI agents that comprises an AI model augmented with users’ data access. AI agents can be connected to a chat interface, like Slack or MS Teams, to create chatbots.

See details following this link for Agents and this link for Chatbots.

Step-by-Step Tutorial

This tutorial demonstrates how to build AI agents with MindsDB using MindsDB SQL editor. This can be also accomplished with APIs and Python SDK.

Let’s list all the steps required to build an AI agent.

1. Create a conversational model

   Create a conversational model using the LangChain integration.

2. Create skills

   Create one or more skills to be assigned to an agent. Note that skills store data to be passed to an agent, so it is required to connect users’ data to MindsDB before creating skills.

3. Create an AI agent

   Create an AI agent providing the conversational model and the set of skills.

4. Create a chatbot

   Optionally, connect an agent to a chat interface to create a chatbot.

The following sections walk you through the process of building an AI agent.

Step 1. Create a conversational model

Use the CREATE MODEL statement below to create a conversational model. If required, adjust the parameters and prompts to fit your use case.

CREATE MODEL conversational_model
PREDICT answer
USING
    engine = 'langchain',
    openai_api_key = 'YOUR_OPENAI_API_KEY_HERE',
    model_name = 'gpt-4',
    mode = 'conversational',
    user_column = 'question' ,
    assistant_column = 'answer',
    max_tokens = 100,
    temperature = 0,
    verbose = True,
    prompt_template = 'Answer the user input in a helpful way';

Ensure that the model status reads complete using this command:

DESCRIBE conversational_model;

Learn more about models created with LangChain.

Step 2. Create skills

A skill is essentially users’ data fed to the model, so the model can answer questions over users’ data.

First, connect your database to MindsDB. Here the sample database is used.

CREATE DATABASE datasource
WITH ENGINE = "postgres",
PARAMETERS = {
    "user": "demo_user",
    "password": "demo_password",
    "host": "samples.mindsdb.com",
    "port": "5432",
    "database": "demo",
    "schema": "demo_data"
};

Create a skill using the connected data.

CREATE SKILL text2sql_skill
USING
    type = 'text2sql',
    database = 'datasource', -- connect your database with CREATE DATABASE and pass its name here
    tables = ['car_sales'], -- list table(s) to be made accessible by an agent
    description = 'this is car sales data';

Note that there are two types of skills: text-to-SQL and knowledge bases. Learn more about skills here.

Verify that the skill has been created successully using this command:

SHOW SKILLS;

Step 3. Create an AI agent

Now that both the conversational model and the skill are ready, let’s create an AI agent.

CREATE AGENT ai_agent
USING
    model = 'conversational_model',
    skills = ['text2sql_skill'];

Verify that the agent has been created successully using this command:

SHOW AGENTS;

At this point, you can query an agent to ask questions over the data.

SELECT question, answer
FROM ai_agent
WHERE question = 'how many cars were sold in 2016?';

Step 4. Create a chatbot

Optionally, you can create a chatbot by connecitng an AI agent to a chat interface.

First connect a chat interface to MindsDB. Here the Slack connection is made.

CREATE DATABASE mindsdb_slack
WITH
  ENGINE = 'slack',
  PARAMETERS = {
      "token": "xoxb-xxx",
      "app_token": "xapp-xxx"
    };

Follow the instructions on how to connect Slack to MindsDB for this use case.

Now create a chatbot providing the AI agent and the Slack connection.

CREATE CHATBOT ai_chatbot
USING
    database = 'mindsdb_slack', -- connect a chat interface with CREATE DATABASE
    agent = 'ai_agent'; -- create an agent with with CREATE AGENT

Verify that the chatbot is running using this command:

SHOW CHATBOTS;

Now you can go ahead and chat with the AI agent via Slack.