Data platforms move toward apps

Snowflake and Databricks are pushing beyond data storage and processing into software for building artificial intelligence apps and live operational tools. (snowflake.com ) (databricks.com) A data platform started as a place to store, clean and query company information; now both vendors are pitching it as the place to build the app itself. Databricks said on April 13 that an official connector with Lovable lets teams build internal tools on enterprise data “without the backend bottleneck.” (databricks.com) Databricks has also been packaging more of the application stack around that pitch. In February, it said Databricks Apps and a new AppKit workflow let developers build and deploy data and artificial intelligence applications directly on Databricks, including from Replit. (databricks.com) Snowflake is making a similar move by wrapping app interfaces around its own models and data services. Snowflake’s developer guides show Streamlit apps running inside Snowflake and calling Cortex agents through the `agent:run` interface, while its Native App framework supports bundling Streamlit into Snowflake-hosted apps. (snowflake.com) (docs.snowflake.com) The technical shift is from dashboards that people check after the fact to software that acts on current data while work is happening. Databricks now describes its developer offering as tools to build “low-latency apps and agents directly on your enterprise data,” and Snowflake’s March 18 preview of Project SnowWork framed the next step as software that can complete multi-step tasks from conversational prompts. (databricks.com) (snowflake.com) That app push is also moving from screens to sensors. RFID Journal reported on April 13 that Wiliot is building its Physical Artificial Intelligence platform and supply-chain automation tools on Databricks so customers can ingest and analyze real-time streams from battery-free Internet of Things tags attached to products and assets. (rfidjournal.com) In plain terms, “physical artificial intelligence” means software making decisions from what is happening in the physical world, not just from spreadsheets or chat logs. Wiliot said its Internet of Things Pixels can turn products into connected data sources, and the Databricks tie-up is aimed at inventory, cold-chain and disruption alerts inside a governed data environment. (rfidjournal.com) Snowflake is using the language of agents just as aggressively. Its 2026 predictions report said enterprise agents will move into “more complex decision paths,” and recent developer tutorials walk users through building multi-tool agents that choose between search and text-to-SQL systems depending on the question. (snowflake.com 1) (snowflake.com 2) The competition is no longer only about whose warehouse runs queries faster or cheaper. It is increasingly about which platform can keep the data, the model, the app interface and the workflow in one place long enough for companies to build software on top of it. (snowflake.com) (databricks.com)