Ask Dorin: Bridging the Gap Between Public Services and Citizens
Inspiration
Our journey at the hackathon began with a vision to improve accessibility to public services. We were inspired by the idea of creating a bridge between citizens and government resources using cutting-edge technology. In an age where information is key, we believed that citizens should have easy access to government documents and information.
Problems we solve
Ask Dorin addresses known problems such as:
Articles use domain-specific language: We simplify the language used in government documents, making them more accessible to the average citizen.
Some articles are too long with irrelevant information: Our AI model extracts the most relevant information from lengthy documents, saving users time by providing concise responses.
Unfriendly user experience: We have created a user-friendly web and mobile app that ensures a smooth and intuitive experience for all users.
What do we bring
With Ask Dorin, we bring:
Search time reduced by at least 60%: Users can find the information they need quickly and efficiently, saving them valuable time.
Natural conversation language responses: Our AI model engages users in natural, human-like conversations, making interactions with government data more approachable.
Easy user experience: We prioritize user-friendliness, ensuring that citizens of all backgrounds can access public services information effortlessly.
What We Learned
Throughout the hackathon, we learned the importance of collaboration and leveraging existing services and tools. We realized the potential of AI in revolutionizing the way citizens interact with government data. We also gained valuable insights into efficient web and mobile app development, deployment, and scaling strategies.
Building Ask Dorin
Ask Dorin is a revolutionary web and mobile app that connects users to public services data through a conversational AI model. We integrated several powerful technologies and services to make this vision a reality:
OpenAI for LLM Models: We harnessed the power of OpenAI's Language Model (LLM) to build our conversational AI. This allowed us to provide users with intelligent and context-aware responses.
Vercel for Front-End Scaling: Vercel played a crucial role in ensuring that our web application could scale seamlessly to accommodate a growing user base. Continuous deployment was made effortless thanks to Vercel.
Google Cloud Run for Backend: Our back-end infrastructure was powered by Google Cloud Run, providing us with continuous deployment, load balancing, and the ability to scale based on incoming requests. This ensured that our service remained responsive even during high demand.
Embeddings and Cognitive Filtering: We didn't stop at just using AI for conversations. We incorporated embeddings for article filtering and extra cognitive filtering based on the retrieved data, ensuring the accuracy and relevance of the information provided.
Public APIs from Portalul de Servicii Publice: To source the article contents, we utilized public APIs provided by Portalul de Servicii Publice, ensuring that our data was up-to-date and aligned with official records.
Challenges Faced
Building Ask Dorin presented its own set of challenges. Some of the key hurdles we encountered included:
Data Accuracy: Ensuring that the information retrieved from public services was accurate and up-to-date required careful handling of data sources.
Scalability: With the potential to serve a large user base, ensuring that our application could scale without compromising performance was a significant challenge.
User Experience: Designing a user-friendly and aesthetically pleasing web and mobile application that catered to both desktop and mobile users required meticulous attention to detail.
Conclusion
Ask Dorin represents the culmination of our efforts at the hackathon. It's a project that aims to simplify and enhance the way citizens access public services information. By bringing together AI, cloud computing, and public data sources, we believe we've taken a significant step toward improving transparency and efficiency in government-citizen interactions. We're proud of what we've achieved in just a short time, and we expect the search time to be reduced by at least 60%.