The Client:
Our customer is an innovative SaaS platform for analytics. It is a fast-growing startup based in the USA. Its vision is to take the data intelligence experience to the next level by
leveraging big data technologies and predictive analytics. They have many customers across different industries who use our product to get insights from their data daily.
Data Analysts: Enhancing Analytics with AI
The Business and Technical Challenges:
Backend of customer's platform developed with Scala.
Our engineers worked on a solution that would allow platform users to easily manage data, make queries, visualise graphs, create forecasts based on existing data, etc:
-
рarse a user request in human language with business terms;
-
form an appropriate query to the database;
-
display the results of the query execution from the database in dashboards;
-
generate SQL from the user's text input;
-
visualise the results.
The Solution:
Our team decided not to use the previous Rasa neural model. Instead, we decided to rewrite the processing of user requests so that they are only in the code: the user's request is processed using AI models from OpenAI.
The Tech Stack Used in the Project:
-
Scala, Python
-
OpenAI API, LangChain
-
Chroma (with VectorDB)
-
RAG pattern
The Result:
Our team of engineers improved the platform as follows:
-
created tools with which platform users can independently and quickly collect their data and visualise it in the required context;
-
added new analytics capabilities;
-
facilitated the work of data analysts in data collection and forecasting.