VeraIB
AI-Powered System for Automotive Intelligence
An AI system that turns plain-language questions into validated warehouse queries. In development.
- Client
- VeraIB
- Year
- Type
- Natural-language BI system
- Role
- AI architecture, data modelling, and product build
- Views
- 461
Context
Why did VeraIB need this?
An automotive group's answers lived in a 35-table warehouse covering vehicle sales, parts and after-sales, reachable only by whoever could write database queries.
Product
What did we build?
A business-intelligence system where a question in plain language becomes a validated SQL query and a streamed answer: narrative, data table, and chart.
Experience
What is it like to use?
Ask in English or Portuguese. The system searches the warehouse, builds and checks its own query, runs it without changing anything, and composes the answer from the data it used.
Engineering
How is it built?
The system runs inside the client's own environment and answers from the data it was given rather than from general knowledge. It can read but never change anything, several independent checks enforce that, and its accuracy is measured against a fixed set of questions so a change that makes it worse is caught before release.
Outcome
What changed?
In active development with the pipeline, safety model and evaluation harness in place. The entry's public copy stays modest until the client ships it.