Memoria AI
Google GenAI & MongoDB Hackathon Project

The AI Assistant That
Remembers and Reflects.

Standard chatbots forget everything once your conversation ends. Memoria AI introduces a persistent long-term memory system. It extracts facts, tracks goals, structures tasks, and reflects on your progress—powered by Gemini and MongoDB Atlas.

Chief of Staff MemoryActive

Semantic Preferences

"Prefers TypeScript, MongoDB, Next.js"

Active Goals

"Acquire 100 SaaS beta customers"

Reflective Insight

"Postponing customer calls; code focus bias"

Live Retrieval Demo

User: "What should I prioritize this week?"

Chief of Staff: "Based on your active goal to acquire 100 beta customers, and my reflection that you have deferred marketing tasks, you should prioritize setting up the email sign-up form over adding codebase features."

Injected 3 semantic memories from MongoDB Atlas using Vector Similarity search

Four Dimensions of Long-Term Memory

Memoria AI categorizes interactions and structures them into specific knowledge schemas inside MongoDB, allowing the agent to reason dynamically.

Episodic & Semantic

Captures life events, milestones, and structural preferences. Remembers details like your tech stack, business metrics, or strategic pivots.

Goal Alignment

Tracks your multi-month objectives. Connects conversations to high-level goals and monitors your development progress automatically.

Task Management

Extracts actionable to-do items from conversations. Helps you schedule work, track completion, and stay accountable.

Reflection Loop

Periodically reviews your memory log, deduces patterns, and publishes high-level behavioral insights to guide future decision-making.