Tables are a local database within your agent’s environment that let you store structured data.

You can use them to store information like customer names and information, a product catalog, or for creating dynamic and personalized interactions.

You can create a new Table directly from this menu. Here, you can add columns, specify your data and schema types, or import a Table directly from a CSV file.

When creating a column, you can mark it as searchable, which tells your agent to treat it as part of its Knowledge Base. Marking a table column as Computed lets you perform AI-powered or programmatic computations on the data contained in a given row or record.

You can add data to a Table by manually creating records, or you can have your agent interact with Tables throughout a conversation or workflow. This can be done programmatically through code or cards, or dynamically by letting an LLM decide when to add, delete, or update records.

Once your Table’s been populated with data, you can use the built-in filtering to search for specific kinds of information.

Tables are also used by functional integrations like the conversation analyzer. Agents that properly use Tables benefit by combining their spreadsheet-style interface with the unstructured data contained in your Knowledge Bases.

‍