Hion and Safetum
Summary
Safetum
Company Size- SME
- Real Estate & Property Management
- Customer Relationship Management (CRM)
- Operations Management
- Product Development & Innovation
- Risk Management & Compliance
Project
Project Types- Customer experience
- 10 001 – 100 000€
- Finland
- Multiple
Implementation
Implementation Level- Production
- Conversational AI & Chatbots
- Natural Language Processing (NLP)
- Azure
- Google Cloud Run
- Google Cloud Vertex AI
- Node.js
- React
- Strapi
Yes
The statutory emergency plan for housing companies has traditionally been a static stack of papers, often ignored by residents and left gathering dust on the shelf. – Dry content: The information is difficult to digest and fails to inspire residents to familiarize themselves with emergency preparedness. –Poor reach: The traditional format does not effectively reach…
Solution
‘Safetum and Hion have developed an interactive, AI-powered service designed to replace traditional paper emergency plans. The heart of the solution is Safetum-Nalle (Safetum Bear), which makes fire safety accessible and understandable for all ages through natural conversation. It transforms complex safety issues into clear answers: the resident asks, and the Bear responds – drawing from either the housing company’s specific data or reliable general sources. Technically, the Safetum Bear is built on the Google Vertex AI platform and leverages the advanced Gemini language model. The system’s intelligence relies on a modern Hybrid RAG architecture, which ensures the accuracy and reliability of answers by seamlessly combining two distinct data sources: Intelligent retrieval and reliability: The system automatically determines whether a resident is asking for general safety information or specifics about their own housing company. Answers to general questions are retrieved from verified official sources without guesswork. For building-specific questions (e.g., the location of the civil defense shelter), the AI responds strictly based on that specific building’s emergency plan. Fact-checking and data security: Critical details, such as maintenance company contact information, are retrieved via the system’s “shortcuts” to ensure 100% accuracy. Data privacy is built-in – private housing company data is never mixed with general information or used to train AI models, guaranteeing complete data hygiene.
Outcome
This innovation has transformed a tedious compliance burden into a concrete safety benefit that serves both housing company administration and residents. Enhanced safety: The barrier to engaging with the emergency plan has been lowered, raising residents’ preparedness and awareness. Convenience: Property managers and boards gain an easy way to fulfill statutory obligations with a modern tool. Accessibility: The service reaches all demographic groups through multilingual support and a clear user interface. Social impact: The solution not only saves time but, thanks to improved preparedness, can save lives in a real emergency.
Contact person:
Jenny Salmimäki
Hion