Creating AI Applications with Postgres: From Semantic Search to Database-Driven Chatbots

AI applications are increasingly important for driving innovation, improving customer service, and accelerating business decision-making. As organisations seek to combine structured data with large language models, there is growing demand for professionals who can design robust, database-driven AI systems.

Designed for experienced developers, architects, and technical decision-makers, this advanced course guides learners through the full process of building an AI application with Postgres. Participants will develop a data model, integrate a PostgreSQL database with an externally hosted large language model (LLM), and combine conventional SQL queries with vector-based retrieval techniques. By the end of the course, learners will have built a working proof of concept for each stage of the process and will be able to identify the technical requirements for developing effective AI applications.

Session 1

- Quick recap of PostgreSQL fundamentals required for the course:

  - Data concepts: tables, rows, columns, data types, operators, and indexes

  - SQL query concepts: SELECT, UPDATE, INSERT, DELETE, and common table expressions

  - Stored procedures: PL/pgSQL and PL/Python

- Differences between AI applications and conventional database applications

Session 2

- Vectors and embeddings as the building blocks of AI applications

Session 3

- Integrating databases and large language models to generate embeddings

- Semantic search as a core building block for AI applications

Session 4

- Combining semantic search with SQL retrieval

Session 5

- Building a working AI chatbot by adding chat completion

At the end of the training, learners will be able to

- design a simple AI application that integrates a PostgreSQL database with a large language model

- generate and use embeddings in Postgres for vector-based retrieval

- implement a semantic search engine using relational database content

- combine SQL retrieval with semantic search in a database-driven AI workflow

- build a chatbot that answers natural-language questions using relational database content

The course uses lectures, real-world examples, practical exercises, and project-based learning. Learners work hands-on with their own PostgreSQL database to model data, connect to a large language model, generate embeddings, and implement an AI application.