music & video
AI & Refugees community support
An AI-powered tool that streamlines
the translation of legal documents for refugees
1st prize
Hack To The Rescue
32 hours
for the MVP
40
teams competing
Challenges
Cultural sensitivity and context
Legal documents often contain specific terms and concepts not directly translatable to other languages. After reviewing the documentation provided by Social Bee, the Miquido team quickly realised that their tool needed to recognise and adapt to cultural nuances. The AI-based application was meant to act as a bridge between refugees and their assistants or law enforcement services in Germany. Therefore, the team had to ensure that translated content retains its intended legal meaning and context.
Optimal performance and responsiveness
Firstly, a dedicated translation tool must perfectly balance the legal intricacies with user-friendly simplicity. Secondly, real-time translation is crucial for time-sensitive legal procedures. Therefore, the team prioritised creating a solution that’s intuitive, easy to use, and accessible to all. Finally, our experts also considered scalability to ensure that refugees have uninterrupted access to vital legal documents during peak usage times.
Language diversity and accuracy
Legal documents are known for their formal language, which can be challenging for those who come from diverse linguistic backgrounds. Accurate translation of these documents is crucial, and it can be a difficult task since it requires handling multiple languages while maintaining high precision and accuracy. To ensure that the translations are reliable and easy to understand, the team had to consider several factors, such as dialects, technical legal terms, and cultural nuances.
Solutions
Product Design Process
Refining the product idea based on user needs
The high-level objective was clear: to develop an AI-powered tool that could aid in the translation of legal documents for refugees. However, our team was determined to create a solution that would perfectly cater to the needs of primary users, filling an existing market gap. To achieve this, we followed a simplified product design process, which included conducting competitive research and analysing our target group’s motivations and pain points. Through this process, we discovered that the product’s success would rely on providing professional assistance with legal matters, helping to overcome language barriers, and offering support in understanding complex information and documents.
Proto-Personas
Adin
32 years old student, who needs to get his university degree get recognised in the country he just have entered. Also, as a refugee, he needs to know which authorities and what documents he needs to get work permit to be able to earn a living.
Sofia
28 year old volunteer, who wants to support every newcomer in the country and provide help related to formal procedures, communication and documentation.
Proof of Concept Approach
Cost-efficient validation of technical solutions
To ensure that the product development plan adhered to technical constraints and the timeline, we utilised a Proof of Concept approach. Firstly, we used a customer journey framework to design a smooth end-to-end user experience while identifying improvement areas. Then, the team employed the ICE growth strategy framework to simplify the product and align its requirements with the hackathon’s tight schedule.
This framework helped us evaluate the different functions of the product based on cost, technical complexity, and user value. In the end, the team was able to identify the most impactful features, determine the final product scope, and deliver the project right on time, with the potential for further scaling.
AI Development
Creating a GPT-3.5-powered chatbot and a translatorÂ
As part of our collaboration with Social Bee, our team developed two products: an online translation tool and a chatbot that helps users understand complex legal concepts. Both products were powered by GPT-3.5 integrations, which allowed for human-like responses and natural language understanding. The chatbot was designed using prompt engineering techniques and acts as a personal advisor to help users understand legal complexities, answer specific questions, and keep track of inquiries and responses. This feature is especially valuable for refugees facing time constraints and stress, as it enables them to receive prompt, professional assistance or revisit their case. Overall, both tools proved extremely helpful for providing accurate legal information to those who need it.
Results
Scalable Proof of Concept
At a hackathon, the team showcased a Proof of Concept that includes crucial AI-powered features. By utilising the OpenAI API and integrating with GPT-3.5, the tool can easily convert PDF files into easy-to-understand documents and provides plain German-English translations, as well as valuable summaries of intricate legal documents.Â
Open access tool that is cheap to maintain
Miquido has developed a Proof of Concept that is designed to evolve dynamically in future iterations. The team plans to include additional features such as document download, sharing, text-to-speech functionality, account management, and document validation in the upcoming product development stages. Initial users, including Social Bee volunteers and refugees, will soon evaluate our Generative AI solution. But even now, it is clear that a simple tool based on the open access software, created within just two days, shows unprecedented potential in optimising internal operations.
Accessibility across web and mobile platforms
When creating digital products for broad audiences, ensuring accessibility is crucial. The solution provided by Miquido offers versatile accessibility features such as text enlargement, light and dark modes, and keyword highlighting. Furthermore, the tool has been adapted for both web and mobile platforms.
