Plain English in, SQL out.

Schema-aware text-to-SQL · mock executionFully client-side

Ask a business question

Type a question or pick an example.

Mock execution preview

Sample schema (3 tables)

customers ( customer_id INT, customer_name VARCHAR(100), region VARCHAR(50), signup_date DATE, total_spent DECIMAL(12,2) )

products ( product_id INT, product_name VARCHAR(100), category VARCHAR(50), price DECIMAL(10,2) )

orders ( order_id INT, customer_id INT, product_id INT, quantity INT, order_date DATE, revenue DECIMAL(12,2) )

What this demonstrates

This demo uses transparent keyword rules rather than a live LLM, on purpose: you can view source and see exactly how intent maps to SQL. In production the parsing is done by a language model grounded in your real schema and governed semantic layer, which is the difference between a toy and a system your finance team can trust. The architecture that makes it trustworthy is the subject of this case study.

Conversational analytics, governed.

AI analytics sprint →