The challenge
A consulting firm needed to improve communication and access to information across its teams without compromising data privacy.
Third-party LLM services were not an option, so the firm needed a chatbot running on its own custom LLM, which would give it control over data privacy, model training and updates.
Our solution
We developed a web application that lets users talk to a custom large language model through a chatbot interface. The result is a private, hosted solution similar to commercial ChatGPT, with additional features.
The interface is deliberately simple: users log in, type a question or statement and receive a response. Each answer comes with its data sources, including references to the documents it was generated from, which makes responses transparent and traceable.
Instead of relying on third-party LLM services, the chatbot runs on a custom LLM developed in-house, giving the firm control over data privacy, training processes and updates.
- Web application with login and a conversational chatbot interface
- Custom LLM developed in-house rather than a third-party LLM service
- Responses with data source details and document references
- Upload of custom files to set the context of a conversation
The outcome
The chatbot changed how the client works with internal knowledge. A straightforward interface and answers linked to their sources made internal communication more transparent and traceable.
Support for uploading custom files to set context allowed the consulting firm to build its own knowledge centres on top of the system.
