search

Data Agent

hashtag
Data Agent

The BDB Data Agent functions as a central controller, autonomously processing, analyzing, and interpreting extensive datasets. By leveraging advanced Agentic AI and pre-configured Large Language Models (LLMs), it identifies trends, detects anomalies, and generates actionable intelligence. Moving beyond conventional dashboards, it facilitates conversational analytics that support proactive decision-making. Its purpose is to convert complex data into accessible, intuitive conversations, thereby enhancing organizational agility and optimizing operational expenditures.The BDB Data Center serves as the central hub for managing and organizing data from diverse sources. It offers native connectors for various databases (SQL, NoSQL), APIs, flat files, and more, ensuring comprehensive data accessibility. Additionally, it hosts "micro functions" that the Data Agent can utilize to suggest and trigger actions.

hashtag
Creating a Data Agent: Overview & Steps

The deployment of intelligent data agent functionality within the BDB platform imposes initial training against the relevant datasets. This training comprises two essential components:

hashtag
Dataset Selection

Using an existing data connector, you can explore your available databases and select relevant tables from your data model. These tables represent the core domain entities and fact/dimension relationships that the agent will interact with.

hashtag
Knowledge Base Upload

Once the tables are selected, you must upload a well-structured knowledge base document. This document serves as the foundational context for the LLM (Large Language Model) that powers the agent. It should include:

  • Descriptions of each table and its purpose

  • Key relationships and join paths

  • Business definitions of metrics and KPIs

  • Supported user intents and how they map to SQL actions or visualizations

  • Any standard business logic or filter conditions

The knowledge base acts as the training material for the agent, enabling it to interpret natural language queries and respond with meaningful insights, visualizations, or actions based on your data.Once these steps are completed:

  • The agent understands the structure and semantics of your data.

  • It can be queried through natural language prompts.

  • It is capable of generating SQL queries, dashboard views, or diagnostic suggestions depending on the use case.

Last updated