17. March 2025 By Anton Schönle and Frederik Julius Szmania
PSLifeGPT - AI meets life insurance
Optimised knowledge management and information searches with the in|sure PSLife chatbot
Tired of wading through endless pages of software documentation? Our new chatbot will do it for you and put an end to the paperwork!
In our blog post, we present an exciting project that represents a business-relevant use case for the application of generative AI. It uses an RAG approach to make the extensive documentation of a comprehensive software product in the insurance industry available in the chat.
What is in|sure PSLife?
in|sure PSLife is adesso's powerful policy management system for the life insurance industry. It includes product modelling, policy management and claims processing and enables the efficient and flexible administration of life insurance contracts. Thanks to many years of use, in|sure PSLife offers numerous advantages:
- Holistic software for contract management
- High degree of automation through integrated dark processing
- Intuitive user interface
- Migration process for large portfolios proven in practice
- Can be operated as an on-premise, cloud or SaaS solution
- Modern Java technology
- Markov approach: Simplification of the tariff landscape using high-performance formulas
- Product modelling: Fast and needs-based modelling of new and existing products
- Contract and service processing: Customisable business transactions throughout the entire contract life cycle
Policy management for life insurers – with adesso and in|sure PSLife
Maximise efficiency and flexibility in life insurance administration! With in|sure PSLife, adesso offers you a powerful policy management software that automates your processes, meets regulatory requirements and guarantees you maximum future-proofing. Benefit from state-of-the-art technology that helps you reduce costs and provide even better customer care.
Find out more and take your policy management to the next level.
What is RAG?
Retrieval Augmented Generation (RAG) is an innovative approach that combines the capabilities of Large Language Models (LLMs) with the advantages of information retrieval. In contrast to traditional methods of text generation, which are based solely on existing training data, RAG integrates external sources of knowledge into the generation process. This is done by searching relevant documents and embedding the found information into the generated responses. This makes the answers more precise, up-to-date and well-founded.
Specifically, the semantic search is used with the help of vector space models, which recognise similar meanings of questions and texts in documents. This makes information available that is not publicly accessible, and the LLM always refers to information from real sources, without the risk of hallucinations.
Project objective and added value
The aim of this project is to develop an intelligent chatbot that searches the in|sure PSLife documentation for user queries and provides qualified answers with correct source information. The chatbot should offer the following added value:
- Improved professional and technical understanding for users
- Support with defining requirements
- Support with creating documentation
- First-level support for questions about in|sure PSLife
- Quality improvement for projects and product management
- Improved maintainability
- Support with test case creation
- Support with release upgrades
- Faster issue management
- Reduction of silo knowledge
- Architecture of the application
The architecture of the application is structured as follows:
- Frontend App UX: A user-friendly interface through which users can enter their requests.
- App Server Orchestrator: Coordinates the various components of the application and forwards the requests.
- Azure AI Search with connected data sources (in|sure PSLife documentation): Searches the connected data sources for relevant information.
- Azure OpenAI: Uses the capabilities of OpenAI's Large Language Model to generate the final answers.
- Cosmo DB: Database for storing chat histories and enabling users to access their own chat histories.
Process flow
The process is multi-stage:
- 1. User prompt: Users ask a question.
- 2. Search Query: The system generates a search query.
- 3. Search Results: The relevant documents are searched and relevant matches identified.
- 4. System Prompt + Search Results + User Prompt: The information from general prompts, search results and user questions are summarised.
- 5. Response to the user: Users receive a qualified response with source information.
Test phase and current status
An extensive internal test phase with PSLife users has been carried out. As things stand, we see the chatbot as an intelligent search engine that provides users with optimal sources from the software documentation in response to their questions. Users can sift through the provided sources themselves to ensure the best possible quality. This massively accelerates time-consuming manual research within the documentation and reduces the need for queries to in|sure PSLife experts. The internal go-live will take place soon so that the advantages gained can be utilised within adesso. At the same time, the application will be further developed to make the added value available to customers as well.
Further AI use cases for the insurance industry
From personalised customer service to automated claims processing, we would be happy to show you specific use cases that optimise your processes, increase efficiency and secure competitive advantages.
