Inspiration
We built EKARA around a specific situation: an older adult living independently at home while their children or caregivers live elsewhere.
For many older adults, everyday tasks like remembering appointments, checking what is happening around the home, or contacting family for help can become unnecessarily difficult when technology requires navigating multiple apps, screens, and interfaces.
We wanted to create an experience where asking for help could be as simple as speaking naturally.
That led to EKARA — Intelligent support for safer, more independent living.
Our goal is not to replace family members or caregivers. Instead, EKARA acts as a bridge between an older adult, their home, and the people they trust.
What it does
EKARA is a voice-first intelligent companion for older adults living independently.
Users can interact with EKARA naturally to:
- 📅 Manage daily routines — Ask what tasks, appointments, or reminders are coming up.
- 🏠 Understand home events — Ask about relevant activity around the home.
- 🆘 Request assistance — Ask EKARA to notify a trusted caregiver or family member.
- 👨👩👧 Stay connected — Help remote family members respond when their support is needed.
For example:
“EKARA, what do I need to do today?”
“Is anyone at the front door?”
“Please let my daughter know that I need help.”
Instead of forcing users to learn a complicated interface, EKARA lets them ask naturally and receive actionable assistance.
How we built it
We built EKARA as a voice-first system connecting Alexa+ with an MCP-powered backend and AWS services.
Our architecture consists of:
- Alexa+ — Natural voice interaction layer
- MCP Server — Connects conversational requests to EKARA's capabilities
- Streamable HTTP — Enables communication with the MCP server
- AWS services — Cloud infrastructure and intelligent processing
- Node.js + TypeScript — Backend and tool layer
- React + TypeScript — Companion dashboard
- Structured tools — For routines, home awareness, and caregiver communication
The MCP layer is the core of EKARA. Instead of allowing the AI to directly perform arbitrary actions, we expose well-defined tools such as:
get_daily_schedule()
check_front_door()
notify_caregiver()
This gives the conversational interface a structured and predictable way to interact with EKARA's functionality.
We built the core interaction flow first, tested the individual tools, and then connected them to the user-facing experience.
Challenges we ran into
One of our biggest challenges was turning natural language into reliable actions.
A request like “I need help” sounds simple, but the system needs to understand the intent, select the appropriate tool, provide the required information, execute the action, and communicate the result clearly.
We also had to work through:
- Designing reliable MCP tools and structured inputs
- Connecting the MCP server through Streamable HTTP
- Handling home-awareness events that may be unavailable or simulated
- Making AI responses predictable rather than overly conversational
- Designing an interface that does not overwhelm older users
- Thinking about privacy, permissions, and failure states
- Connecting multiple technologies while keeping the architecture maintainable
These challenges pushed us to think beyond the AI model itself and focus on reliability, accessibility, and user trust.
Accomplishments that we're proud of
We're proud that EKARA brings several technologies together around a specific human problem rather than building another generic AI assistant.
Some of our biggest accomplishments are:
- 🎙️ Building a voice-first experience designed around natural interaction.
- 🔌 Connecting Alexa+ to a custom MCP-based tool layer.
- ☁️ Integrating AWS capabilities into the application.
- 🧩 Designing modular tools that can be extended with new capabilities.
- ♿ Keeping accessibility and simplicity at the center of the experience.
- 👨👩👧 Creating a clear connection between independent living and remote caregiver support.
- 🚀 Building the foundation so EKARA can evolve into a broader home-assistance platform.
Most importantly, we created a system where AI is not the destination — it is the interface that makes useful actions easier to access.
What we learned
EKARA taught us that building an AI product is much more than connecting an LLM to an interface.
We learned how important it is to design the entire interaction pipeline:
User → Voice → AI → Tool → Action → Feedback
We also learned how MCP can provide a structured bridge between conversational AI and real application capabilities.
Beyond the technical side, EKARA changed how we think about accessibility. A good accessibility experience is not necessarily about adding more features — sometimes it means removing unnecessary complexity.
We also learned to design for failure. When an external service is unavailable or an action cannot be completed, the system needs to communicate that clearly rather than pretending everything worked.
What's next for EKARA
Our next goal is to make EKARA more capable while keeping the experience simple.
We want to explore:
- 🏠 Deeper smart-home and Ring integrations
- 🧠 Personalized routines based on user preferences
- 🔔 Proactive reminders and assistance
- 👨👩👧 A richer caregiver dashboard
- 🔐 Stronger privacy and permission controls
- 🌐 Multilingual voice interactions
- ♿ More accessibility-focused features
- 🤝 Additional MCP tools for everyday tasks
In the long term, we envision EKARA becoming a trusted digital layer between people, their homes, and their support networks.
Because independent living should not mean being disconnected.
Log in or sign up for Devpost to join the conversation.