Brain4Data boosts customer service with Oracle Database 23ai and Generative AI
The startup helps small businesses tap the power of AI by relying on Oracle Database 23ai for vector search and application development.
“After only a few months using OCI’s Oracle Database 23ai and Generative AI, our AI agent could answer 80% of end-user questions.”
Business challenges
Have a question? Fred will find you the answer. And Fred will help your customers with their problems.
Fred is the GenAI-powered agent developed by Brain4Data, a German startup focused on combining artificial intelligence (AI) with robotic process automation for small and midsize businesses to make recommendations for their employees or customers. Need to find a driver for a delivery run? Fred scours employees’ schedules and locations and finds the best match in seconds. Want to help sweeten the terms of a business contract? Fred provides information that helps you optimize the sale.
To provide specific information and context that a human employee would have, Brain4Data needed to develop a GenAI agent that uses Large Language Model capabilities, such as text generation, based on the specific knowledge scattered throughout a customer’s documents, databases, and ERP systems to fine-tune the agent’s responses. Brain4Data needed to use retrieval augmented generation (RAG) to access those reams of a customer’s internal data that can be in any format to generate trustworthy answers and actions, while keeping Fred fast, affordable, scalable, and reliable.
Brain4Data’s approach was to abandon the constraints of structured data wherever feasible and turn as much as possible to PDFs. The PDFs become raw material for the RAG system to digest and search for answers to customer or employee questions. At first, the company stored customers’ data in a separate vector database on an external server, but that added complexity and slowed performance. Brain4Data needed a better approach.
Why Brain4Data chose Oracle
Jan-Peter Meyer, who joined Brain4Data as CTO, chose to develop Brain4Data’s technology on Oracle Database because of the data simplicity it provides his team. The company was already using Oracle Database as a document repository, and recognized how the native AI Vector Search capabilities of Oracle Database 23ai simplified the semantic search portion of RAG.
By storing multiple types of data and vector representations together in the same Oracle Database, Brain4Data could simplify both operations and development by managing fewer databases, having fewer data copies, and moving data less. For example, a Fred customer’s original documents stored in Oracle Database 23ai could be turned into vectors that are stored and searched in the same database, helping identify the relevant sections to use when crafting an answer with OCI’s Generative AI for RAG.
Additionally, Brain4Data decided to develop on Oracle Database with APEX and SQL Developer Data Modeler because of the data simplicity they provided the company’s developers, data scientists, data architects, and analysts.
Results
Within a few months of starting to use Oracle Database 23ai’s vector capabilities with OCI Generative AI Service, Fred handled 80% of end-user requests, which is the goal Brain4Data had targeted. In many situations the AI took actions itself based on the necessary context. In early 2024, Brain4Data won the European Seal of Excellence, a prestigious honor awarded by the European Innovation Council (EIC) of the European Union.
Brain4Data uses Oracle APEX, Oracle’s low-code programming environment, to create a rich graphical user experience tailored for each customer. Whether it’s creating a mobile app for an end customer to ask Fred questions, or a browser app that a client’s staff uses to monitor Fred’s actions and operations, Oracle APEX is Brain4Data’s go-to interface design and development tool.
About the customer
Brain4Data is a startup in Germany that uses AI and robotics to help customers streamline workflows and promote logistics efficiencies. The company’s augmented intelligence assistant, Fred, was first released in January 2022.