HackBack AI

Inspiration

HackBack AI was inspired by voice assistants like Alexa, but built with a deeper purpose: remembering everyday personal information when you actually need it. I realized that while smart speakers can answer general questions, they don’t truly remember all information provided to them, and they aren't as portable as a mobile device.

From doctor appointments and pharmacy prescriptions to presentations, small daily reminders, and simple things like where you left your glasses, these moments matter. Memory can be fragile, and important details are often forgotten. I wanted to create something that acts like a personal cognitive backup system, a safe place where everyday information is stored and retrieved naturally through conversation.

What it does

HackBack AI securely stores, organizes, and retrieves your personal memories using natural speech.

You simply talk, and it remembers.

How we built it

HackBack AI, a voice-enabled memory assistant designed specifically for the elderly, is built using React + TypeScript + Tailwind CSS as the frontend stack, coupled with the Web Speech API to facilitate browser-based speech recognition. As the user talks, the speech transcript is sent to a serverless edge function, which invokes the Amazon Nova API to identify the user's intent, whether to store a new memory or retrieve an existing one using the tool-calling capabilities of the AI model. The memories are persisted to a cloud database, which has a memories table with fields to store the text, category, and timestamp of the memories. The edge function invokes the API to retrieve existing memories to provide context to the AI to map recall queries against the persisted memories before responding with a warm speech or storing the new memory.

Challenges we ran into

One of the biggest challenges was balancing the two aspects of privacy and intelligence. Personal memories hold sensitive information, hence the need to provide a secure mechanism to store the data while still enabling AI to retrieve the information was a challenge.

Accomplishments that we're proud of

• Creating a true personal AI memory assistant
• Designing a frictionless, voice-first experience
• Building a secure memory structure system
• Launching a free-to-start model with ad removal options

We’re also especially proud that our product, HackBack AI, helps solve an actual problem that humans face: forgetting the little, important things.

What we learned

We discovered that memory was an emotional thing. It’s not just data; it’s reassurance.

We discovered that people don’t just want their information stored; they want to know it’s there when they need it.

We discovered that simplicity was key. The less they have to think about it, the more powerful it becomes.

What's next for HackBack AI

Next, we plan to:

• Add multilingual abilities
• Add proactive reminders based on memory patterns
• Introduce secure family-sharing options
• Improve on-device privacy features
  

Built With

• amazon-nova-2
• css
• react
• tailwind
• typescript
• web-speech-api