Inspiration
A broken streetlight. A damaged road. Overflowing waste. A water or sanitation problem.
These may seem like small issues individually, but for citizens, they can have a real impact on everyday life. The challenge is often not only identifying the problem — it is knowing where to report it, what information to provide, and how to follow up afterward.
That led us to a simple question:
What if reporting a city problem could be as easy as having a conversation?
We came up with SmartCity AI Assistant, an AI-powered platform designed to help citizens report urban issues in natural language. Instead of navigating complicated forms or trying to identify the correct department, a citizen can simply explain what is wrong.
The assistant then turns that conversation into a structured request that can be processed and routed through the appropriate workflow.
What it does
SmartCity AI Assistant acts as an intelligent layer between citizens and city-service workflows.
A citizen can describe an issue in their own words, and the assistant is designed to:
- Understand what the citizen is reporting
- Identify and categorize the issue
- Extract important details such as location, description, and urgency
- Ask follow-up questions when information is missing
- Create a structured service request
- Route the request to the appropriate department or workflow
- Store the request for tracking
- Trigger notifications and workflow actions
- Help users follow the status of their request
The idea is simple: make reporting easier for citizens while making requests more structured and actionable for service providers.
How we built it
We are building the prototype around an AI agent, automation workflows, APIs, and a structured backend.
The core flow is:
Citizen → AI Assistant → Understand & Classify → Validate Information → Route Request → Store Data → Notify → Track Status
The AI agent handles natural-language understanding, classification, and information extraction. Automation workflows handle the operational side — connecting services, processing requests, updating records, and triggering notifications.
We are keeping the architecture modular so that additional city services and workflows can be added without rebuilding the entire system.
Challenges we ran into
One of the biggest challenges we expect to face is understanding how differently people describe the same problem.
For example, someone might say:
“The light outside my street has stopped working.”
Another person might report:
“There is a broken streetlight near my house.”
The system needs to recognize that both descriptions may refer to the same type of issue while still collecting the information needed to process the request.
Other challenges we are focusing on include:
- Collecting enough information without making reporting frustrating
- Correctly classifying different types of urban issues
- Validating locations and other important details
- Routing requests to the correct workflow
- Handling incomplete or ambiguous descriptions
- Designing the system so it can grow beyond a single type of city service
Accomplishments that we're proud of
At this stage, we are proud of turning a familiar city-service problem into a concrete AI and automation solution that can be developed into a working prototype.
We especially like the idea of connecting AI to real workflows rather than stopping at a chatbot.
The goal is not simply for an AI to answer a citizen's question. It should understand a request, structure the information, trigger the appropriate action, and help move that request through the process.
We are looking forward to validating this approach during the hackathon and seeing how far we can take the prototype with a collaborative team.
What we learned
Even before completing the prototype, the project has highlighted an important lesson: a useful AI system is more than an LLM.
A reliable solution also needs good data handling, validation, tool and API integration, workflow logic, error handling, and a simple user experience.
We are also exploring how AI agents and automation can work together — with AI handling natural-language understanding and decision-making, while deterministic workflows handle actions that need to be reliable and repeatable.
The hackathon will also give us an opportunity to learn how to take a broad urban problem, turn it into a focused product, and build it collaboratively.
What's next for SmartCity AI Assistant
Our immediate goal is to build and demonstrate a working prototype that takes a citizen from initial report to structured request and automated workflow.
From there, we would like to explore:
- 📍 Location-aware issue reporting
- 📷 Image-based issue detection
- 🎙️ Voice-based reporting
- 📊 Service-provider analytics dashboards
- 🔔 Real-time notifications and status updates
- 🌐 Multilingual support
- 🔗 Integration with existing municipal systems
- 📈 More city-service categories
Ultimately, we want to explore how AI and automation can make city services simpler to access, easier to manage, and more responsive to the people who use them.
Built With
- ai-agents
- artificial-intelligence
- google-maps-platform
- llms
- n8n
- openai-api
- postgresql
- python
- rest-apis
- supabase
- webhooks
- workflow
