Give agents the memory and knowledge they need to reason well, using durable conversation memory and a searchable model of your schema.
Agent memory
Agent memory gives AI agents a durable, per-user memory of past conversations, stored in Postgres and built into the AIDB extension, with per-actor scoping, session and agent filters, safe retries, and auditable deletion.
Install AIDB and configure the Agent Memory Server before using agent memory — see Installing agent memory and Configuring agent memory.
Add, search, list, and delete memories with the aidb_memory SQL functions — see Managing agent memories.
Semantic knowledge bases
Semantic knowledge bases index your database's schema, including tables, views, columns, and comments, into a searchable vector store, so an agent, or you, can find the right relations and columns by meaning instead of guessing from column names.
Install and configure the AIDB extension before creating a semantic knowledge base — see Installing and configuring AIDB.
Create a semantic knowledge base with aidb.create_semantic_kb(), then search it by meaning or use it for text-to-SQL — see Getting started, Searching a semantic knowledge base, and Text-to-SQL.