A woman working at a laptop, with floating panels beside her showing a database linked to tables, a SQL query, a bar chart and a donut chart.

Retail analytics. AI Systems, Data.

A 2.5-million-row warehouse learns to answer in plain English.

2.5m
rows, asked in plain English

2.5m
rows, asked in plain English

Client: Solinovo build, Pakistan
Industry: Retail analytics
What we built: AI Systems, Data
Timeline: V1 complete, 2026

Before

A read-only AI analyst over a PostgreSQL star schema that answers business questions in plain English and shows its SQL.

A data warehouse only answers questions for the people who can write SQL against it. Everyone else joins a queue for a report, and by the time the report arrives the question has usually moved on.

What we built

01
The decision that shaped it
The failure mode of natural-language querying is confident nonsense, so we built for verifiability first. A semantic layer maps business vocabulary to the actual tables, and the generated SQL is always shown in full. An analyst can check the query before trusting the number. Access is read-only by construction, so the worst case is a wrong answer, never a damaged warehouse.
02
What we engineered
A read-only analyst sitting over a PostgreSQL Kimball star schema, with a semantic layer between the question and the query. Results are charted automatically, and the SQL behind every answer is available on the same screen.
03
Plain-English querying over a star schema
Semantic layer mapping business terms to tables. Generated SQL shown in full for verification.
04
What leaves the system
The reporting queue stops being the bottleneck between a question and an answer. Analysts keep their audit trail, because nothing is returned that can't be checked.

01
The decision that shaped it
The failure mode of natural-language querying is confident nonsense, so we built for verifiability first. A semantic layer maps business vocabulary to the actual tables, and the generated SQL is always shown in full. An analyst can check the query before trusting the number. Access is read-only by construction, so the worst case is a wrong answer, never a damaged warehouse.
02
What we engineered
A read-only analyst sitting over a PostgreSQL Kimball star schema, with a semantic layer between the question and the query. Results are charted automatically, and the SQL behind every answer is available on the same screen.
03
Plain-English querying over a star schema
Semantic layer mapping business terms to tables. Generated SQL shown in full for verification.
04
What leaves the system
The reporting queue stops being the bottleneck between a question and an answer. Analysts keep their audit trail, because nothing is returned that can't be checked.

Plain-English querying over a star schema
Semantic layer mapping business terms to tables
Generated SQL shown in full for verification
Automated charting of results
Read-only access across a 2.5-million-row warehouse

What changed

Anyone can ask a 2.5-million-row warehouse a business question in plain English.

The reporting queue stops being the bottleneck between a question and an answer. Analysts keep their audit trail, because nothing is returned that can't be checked.