๐ Inspiration Managing rental properties is time-consuming, stressful, and full of repetitive tasks โ especially for landlords with listings across platforms like Airbnb and Booking.com. We saw a massive opportunity: what if an AI agent could handle everything a property manager does โ instantly, autonomously, and with a human touch? Thatโs how TUMai Lagent was born โ to transform property management into a fully automated, intelligent experience.
๐ค What it does
- TUMai Lagent is an end-to-end property management platform powered by agentic AI that integrates with major platforms like Airbnb and Booking.com. It autonomously:
- Responds to guest inquiries in real-time, even during emergencies
- Calls plumbers, handymen, and cleaners when needed
- Orders replacement furniture or suggests buying options in-app
- Knows every detail about the property and its neighborhood
- Syncs all communications into a unified inbox
- Provides 24/7 intelligent support that feels human and personal
- Itโs like having a full-time property manager โ minus the overhead.
๐ ๏ธ How we built it
- Our custom agentic AI stack uses a combination of LLMs and action-routing pipelines to interpret tenant requests and autonomously take actions like calling services or ordering replacements.
- We layered in property-specific context memory and retrieval-based QA for hyper-personalized answers.
- The unified inbox was built using React and Superbase, syncing all platforms in real-time.
- Emergency-handling logic connects with service APIs (e.g. Twilio, Zapier, Amazon) to trigger real-world actions like phone calls or orders.
โ ๏ธ Challenges we ran into
- Designing AI agents that donโt just chat but act safely and reliably.
- Building a scalable system that can manage real-world consequences (e.g. sending a plumber, ordering the right sofa).
- Integrating multiple rental platforms into one seamless interface.
- Training AI to understand the nuances of guest emotions, urgency, and local context.
๐ Accomplishments that we're proud of
- Developed a fully functional prototype enabling centralized property management
- Deployed Agentic AI v1, capable of handling guest inquiries, emergencies, and automated service calls within a test environment
- Simulated real-world scenarios like broken appliances, emergency plumbing, and replacement ordering โ all handled autonomously by the AI
- Built a unified inbox prototype that consolidates guest communications across platforms into one sleek interface
- Designed and validated advanced agent modules for future expansion, including property knowledge, sentiment detection, and neighborhood intelligence
๐ What we learned
- AI agents can go far beyond chat โ if you design them to act with autonomy and real-world impact.
- Landlords deeply value speed, trust, and zero effort โ and will switch tools for a solution that delivers those.
- Simple UX + powerful automation = massive market potential.
๐ฎ What's next for TUMai Lagent Launching in 3 key European cities this year
Expanding our agent marketplace with plug-and-play modules: โ Legal compliance โ Predictive maintenance โ Multilingual onboarding โ Reputation/review optimization
Partnering with PMS providers and scaling referral-based growth
Building toward our vision of 100M+ stays managed by AI
Built With
- amazon-web-services
- fastapi
- figma
- mistral
- python
- react
- superbase
- tailwindcss

Log in or sign up for Devpost to join the conversation.