Lahwita

Inspiration

Accessing accurate, up-to-date legal information for landlords and tenants in Ontario is surprisingly difficult. By making this information readily available through a generative AI legal assistant, we can deliver significant value to both parties navigating the complex Residential Tenancies system.

What it does

Lahwita is a generative AI assistant specialized in Ontario Landlord and Tenant Board (LTB) matters. It retrieves and reasons over official sources—including the Residential Tenancies Act, 2006, the Statutory Powers Procedure Act, LTB Rules, Practice Directions, Guidelines, and relevant case law—then acts as an informed legal consultant, guiding landlords and tenants toward the correct procedures and approaches for their specific situations.

How we built it

We implemented a clean, layered architecture: (i) a modern chatbot-like frontend UI for intuitive user interaction with the model in a conversational fashion; (ii) a Django backend handling session management, user data, forms metadata, and storage orchestration, and bridging the gap between the frontend UI/UX and the FastAPI; and (iii) a FastAPI-powered layer handling API calls to the Gemini 3 model.

The services we have specifically used for our application are:

Google AI Studio
Pinecone for vectorized indexer database
Steamlit / React for frontend UI/UX
AI-powered FastAPI handling inbound and outbound requests to the Gemini 3 API
Django framework for bridging the gap between the frontend and the AI-powered FastAPI backend

Challenges we ran into

Many LTB application forms are password-protected PDFs that restricted our ability to fully automate form-filling within the hackathon timeframe. While we couldn’t implement complete end-to-end form population this time, we’re working to unlock this feature in future iterations.

Accomplishments that we're proud of

Lightning-fast team formation: The Yonge St. Developers came about from 9 individual developers who convened by accident at the Shopify Builder Sundays event at the heart of downtown Toronto, Canada, on January 11, 2026—just weeks before the Gemini 3 Hackathon deadline. We transformed a group of complete strangers into a cohesive, high-performing team in record time.

Rapid onboarding and role alignment: In under a month, we quickly got to know each other's strengths, built trust, and assigned meaningful contributions that played to everyone's capabilities—turning raw enthusiasm into real project momentum.

Impressive progress under pressure: Despite the extremely short timeline and starting as total strangers, we've delivered a fully functional, multi-layered GenAI application (UI + Django backend + FastAPI AI layer with Gemini-powered RAG) that tackles a genuinely useful problem in Ontario landlord-tenant law.

What we learned

We discovered that building a highly dedicated team from complete strangers and delivering under intense, fast-approaching deadlines can feel overwhelming—especially when no one yet knows each other’s strengths or working styles. Yet with genuine commitment, open communication, and relentless determination, it’s remarkable how much progress a group can achieve in such a short window. In just a few weeks, we went from zero to a functioning, multi-layered GenAI application powered by Gemini 3. This experience has reinforced our belief in the power of rapid iteration, quick learning, and collective momentum. We’re excited to carry this energy, adaptability, and collaborative spirit forward into future projects and continue pushing the boundaries of what we can build together.

What's next for Lahwita

Building on the strong foundation of our Gemini-powered legal AI assistant, we’re excited to roll out several high-impact features in upcoming iterations. Our top priority is enabling intelligent, end-to-end LTB form filling: the system will auto-populate official application forms based on user conversations and retrieved legal context (once we've successfully circumvented the password-protection layer of forms). We also plan to add powerful drafting capabilities, allowing Lahwita to generate well-reasoned legal submission letters, evidence summaries, and other supporting documents drawn directly from the Residential Tenancies Act, LTB Rules, Guidelines, Practice Directions, and relevant case law.