We are back with another wild launch. This time we pulled out the Strata Semantic Layer and stood it up as a standalone service. Its crazy fast and empowers you build your dream data app on top. Want an incident investigator, an embedded dashboard creator, a visual explorer, an alert driven agent loop, or something entirely different? You can now leverage 0sql in your own custom app. Add reliable deterministic SQL queries to your agents.
We never see your data or connect to your warehouse. 0sql does what its good at: generating highly competent, accurate, reliable, and secure SQL queries.
Here is a fantastic demo app that simulates conversational analytics over a customer service domain:
Clone the repo and see how easy it is to setup and deploy a semantic model. Its equally as easy to get your clankers to build something custom with it.
git clone https://github.com/stratasite/0sql-dashboard-demo
cd 0sql-dashboard-demo
npm install
npm run seedInstall zsql, our amazing CLI:
curl -fsSL https://0sql.io/install.sh | shThen deploy the semantic model to 0sql:
zsql auth --project semantic --api-key zsk_… --server https://app.0sql.io
npm run deploy:modelYou should be able to chat with your new app on http://localhost:3000/.
How do we compare?
0Sql is designed for maximum performance and expressiveness. If you can find a more capable semantic layer, let me know. Here are key highlights compared to cube and MetricFlow (dbt).
Automatic Data Blending
Combine data across many fact grains in a single query safely. 0Sql aggregates to the common dimension and combines them. There is no custom authoring of explores, views, or other manual per use case stitching. This works out of box using naming conventions and a special property we call extended blending groups. It’s so easy it feels like magic.
Semantic Routing
Stack several computing engines to deliver the best performance for your analytics. Put last 6 months rolling in an in memory OLAP engine. Leave history in Trino. Optimize cost and performance with 0Sql. Our semantic engine will route queries to the best source that can accurately resolve the query. This is compute aware, partition aware, and aggregate aware semantic querying. Now you can deploy your own OLAP tier you control.
Maximum Expressiveness
From a simple concise YAML based model, 0Sql delivers a highly expressive semantic layer. Out of the box support for snapshot measures, cross domain measures, exclusion (LOD) measures, and inclusion measures. On top of that at runtime your agents or users can create calculations safely across the entire model. Finally, 0sql comes with an easy way to cohort measures or queries. Need to know how signups in january this year compares to signups last year, use segments.
Row Level Security Across the Stack
Not only can we provide row level security on a single datasource, the same security can be applied across the entire stack of compute engines. Same rules on your OLAP engine as in your deep storage tier. One security model deployed every where.
There is a whole lot more check out this comparison table.
Pricing and Monetization
Honest answer, we don’t know yet. We are not sure if a better alternative to Cube and MetricFlow is needed or viable. It’s free for now. We will give you ample notice if and when we figure out pricing.