Empowering vulnerable employees with technology
Project Overview
The goal of this project was to explore how AI and machine learning could be leveraged to improve the work experience of employees at Samhall, a Swedish organization that employs individuals facing barriers to the job market, including cognitive, mental, and physical disabilities.
The focus was on addressing operational challenges during a two-year work placement phase, where employees undertake various roles across different sectors.
Research Approach
- Initial Challenges: Limited access to employees and customers meant we had to rely heavily on desk research. We interviewed a special needs teacher, which helped shape the direction of our research and insights.
- Pivoting Focus: Initially, our team explored AI-driven automation tools for Samhall's operations. However, midway through the project, a client check-in refocused our efforts on designing a chatbot specifically aimed at supporting employees in their daily tasks. The idea was to create an employee-centered tool to streamline processes such as scheduling and guidance, ultimately reducing repetitive tasks for managers by making their roles more human-centric.
Ideation & Development
- AI and No-Code Tools: Our team attempted to build a functional chatbot prototype using Claude and custom prompts via ChatGPT to generate the required databases. However, the complexity of managing four databases and integrating an API became overwhelming, leading us to test a no-code tool (Landbot). Unfortunately, the chatbot responses were unreliable, and the decision was made not to proceed with further investment in that tool.
- Refining the Solution: After pivoting, the team focused on creating personas, user journeys, and chat interactions using generative AI tools such as ChatGPT, Claude, and Midjourney. While the personas were essential for guiding the design, they were intended for internal use only and were not based on real user data.
The chatbot interactions were intentionally limited to simple, structured tasks such as scheduling, transportation guidance, and breaks, ensuring that it did not venture into sensitive areas like conflict resolution or personal information processing were human support would and should always conquer.
Outcome
While the chatbot prototype faced technical challenges, the project provided valuable insights into the potential of AI tools in supporting vulnerable employees.
The final solution will empower employees by providing a reliable, easy-to-use digital assistant and freeing up time from managers to focus on human-centered tasks.