Skip to main content

Understand External Databases

Using an external database in GlobalAI offers significant advantages, especially when you integrate with a client's infrastructure and manage the knowledge base of AI agents.

Connecting to existing knowledge bases​

  • Seamless Infrastructure Integration: Clients often provide their own vector databases, which can be cloud-based or on-premise. AI agents connect directly to these databases, ensuring seamless integration with the client's existing infrastructure.
  • Leveraging Client's Choice: The platform connects to the client's existing databases of choice, allowing AI agents to use them as a knowledge base.
  • Vector Database Support: You can connect to existing vector databases, such as Qdrant, enabling AI agents to retrieve relevant data for their responses.

Contrast with temporary databases​

It's important to note that GlobalAI distinguishes between existing (external or persistent) databases and temporary databases.

  • Permanent (External) Databases: When configuring an agent's knowledge base, users can add a "base knowledge base" by connecting to an existing data vector.
  • Temporary (Internal/Ad Hoc) Databases: In contrast, the system also supports temporary file-based databases, such as ChromaDB or LanceDB. You use these databases for ad hoc knowledge during a single execution. They work like an SQLite database and help you analyze new documents (for example, PDF files or URLs) for the current session only.
Temporary vs. permanent databases

The main advantage of using an external database is continuity and direct access to the client's established knowledge repository. This avoids the need to reprocess or re-index data for each execution and ensures compatibility with the database tools the client already prefers or owns.

Connection configuration​

When configuring an agent's knowledge base to use an existing vector database (like Qdrant), configure the collection, the embedding you use, and the service address.

Additionally, the AI agent connects to these databases to retrieve relevant data for responses, supporting document types such as CSV, DOCX, PDF, and web pages. By default, the system retrieves five documents. You can adjust this number based on the document type, indexing method, and user needs.

Now that you understand the concept of external databases in GlobalAI, you can explore more about how to configure and use them in your workflows.

Connect an External Database

Learn how to set up and connect external databases to your AI agents in GlobalAI.

Manage Serverless Functions

Explore the steps to execute discrete, reusable operations within your automation workflows.