Most production AI applications today are stitched together from several separate systems — a transactional database, an analytical warehouse, a vector store, a search index, and an external inference layer. Every connection between them adds freshness lag, glue code, and operational overhead.

This webinar is an introduction to the Aura AI Platform, SingleStore's platform for building and running intelligent services on top of SingleStore Helios. Bharath Swamy, Senior PM for the AI Platform at SingleStore, will set the context — what is hard about building AI apps on today's stack — then walk through what Aura is, the components it includes, and the kinds of problems each one is designed to solve. The session closes with a short live demo so you can see one of those pieces in action end to end.

The goal is not to make you an expert on AI in 60 minutes. It is to give you a clear picture of what exists, what it is for, and where it might fit alongside the systems you already run.

What You’ll Learn:

•  The problem space — why a five-layer data and AI stack creates freshness, cost, and complexity issues, and what AI workloads actually need from a database.

•  Where SingleStore fits, and where it doesn't. A straightforward look at the workloads SingleStore is built for, and the ones better served by other tools.

•  A tour of what is in the Aura AI Platform today, with the problem each piece is designed to solve:

    • Aura Container Service — serverless compute that runs inside the database VPC

    • Notebooks and Scheduled Jobs — Jupyter-native, pre-authenticated, versioned

    • Cloud Functions — turn any notebook into a secure REST endpoint

    • Python UDFs — call Python directly from SQL

    • Models and InferenceAPIs — one gateway for managed models (Amazon Bedrock, Azure AI) and Aura-hosted open-source models (vLLM, TEI), all behind one credential

    • AI Functions — call a foundation model from SQL for sentiment, summarization, classification, extraction, embeddings, and generation

    • ML Functions — train classical ML next to the data and predict from SQL (classification, forecasting, anomaly detection)

    • Performance Tuning Agent — agent-assisted query optimization, exposed both in the Portal and as an MCP tool

    • Data Migration Agent and Flow — discover, map, validate, and switch over from Oracle, SQL Server, MySQL, Postgres, Snowflake, MongoDB, with schema-drift handling built in

•  A short intro to MCP (Model Context Protocol), and a look at the open-source SingleStore MCP server (20+ tools across data, compute, and platform operations).

•  Live demo — a quick end-to-end walkthrough: building a small custom MCP server in a SingleStore Notebook, deploying it as a Cloud Function, and calling it from Claude or any MCP client. A runnable notebook ships with the session if you want to try it later.

Who Should Attend:

•  Engineers, architects, and technical leaders who want a quick lay of the land on the Aura AI Platform

•  Teams currently building (or planning) AI-powered applications who want to know what is in the SingleStore toolbox

•  Anyone curious about MCP and how AI agents can be wired up to a database

•  Existing SingleStore users who want a single overview of everything Aura now includes

This is an introductory session. No prior Singlestore experience expected.

Agenda (~60 minutes)

1. The Modern Database Landscape — the problems that motivate a platform like Aura

2. Where SingleStore Fits (and where it doesn't) — the workloads it is built for, and the ones better served elsewhere

3. The Aura AI Platform — a tour of the building blocks and the first-party AI services on top of them

4. Live MCP Demo and Q&A — seeing one workflow end to end

Speakers