“Top 10 best-selling products this month”
You type your question to Kai in everyday language. Kai knows your database schema; it turns your question into a safe, parameterised SQL query, runs it and brings the result to you as a table or chart. The query it generates is always visible — it never hides what it runs.
Queries run read-only and are validated; no statement generated by the AI can change anything in your database.
Where does the name come from?
Kai was born from combining Kaizen (continuous improvement) and Artificial Intelligence. The idea of pairing small but continuous improvements with the speed of AI lives in the product’s name. For the full story: Powered by AI.
The real difference: the AI is in your building
With cloud-based AI tools, your question, your database schema and often your data itself go to a provider’s servers. With Kai, they don’t.
Kai’s language engine can run inside your organisation on NVIDIA Jetson + TensorRT or Ollama. No internet connection is required and your questions never leak outside. For data protection (KVKK/GDPR) and commercial confidentiality, this is an assurance cloud solutions cannot give. If you prefer to use a cloud LLM, that option is open too — the choice is yours.
The AI analyst
Kai doesn’t just generate queries. In the report commentary flow it works like an analyst: it interprets the view, examines the data, compares periods and highlights the points that need attention. With responses that stream step by step, you can follow how the result was reached.
Kai works together with ErpwareBI; reporting and AI are subject to the same security rules — role authorisation and row-level security.
Read on
- AI Glossary Text-to-SQL, RAG, on-premises models, inference cost
- What Is Discovery For? If the Machine Finds the Answer, Who Learns the Question? If the machine finds the answer, who learns the question?