Snowflake started as a cloud data warehouse. Today it is a full data platform: it ingests data in near real time, runs Python and AI workloads next to the data, and hosts the applications that end users work with. The posts below give a short overview of three of those areas and what a first step into each of them looks like. They are written for teams and decision makers who want to know what is possible, not for engineers looking for a step-by-step guide.
Nightly batch loads are no longer enough when operations, customer service, and fraud detection need data that is minutes old. Snowpipe, Snowpipe Streaming, Dynamic Tables, and Openflow let Snowflake take in events as they happen and keep derived tables fresh without a separate streaming platform.
Read moreSnowpark brings Python, Java, and Scala to the data instead of moving the data to the code. Cortex adds large language models, document extraction, search, and natural-language analytics as SQL functions and managed services. Together they let a company use AI on its own data without the data ever leaving its governance boundary.
Read moreData is only valuable when somebody can act on it. Streamlit in Snowflake turns a Python script into an internal web app, Snowpark Container Services runs full services next to the data, and Snowflake Postgres and Hybrid Tables bring transactional PostgreSQL-style workloads into the same platform.
Read moreEvery one of these areas can be piloted in a few weeks on top of an existing Snowflake account. If you want a second opinion on which one gives your organisation the most value first, get in touch.