Memo AI - Persistent Memory Agent

Inspiration

What it does

How I built it

Challenges I ran into

Accomplishments that I'm proud of

What I learned

What's next for Memo AI - Persistent Memory Agent

Inspiration

AI agents need persistent memory to be useful in production. CockroachDB provides the always-on distributed foundation.

What it does

Memo AI is an intelligent chat agent with persistent memory. It stores every conversation in CockroachDB with vector embeddings (pgvector), retrieves past memories using semantic search, and generates context-aware responses via AWS Bedrock.

How we built it

Node.js + Express backend, CockroachDB with pgvector extension, AWS Bedrock for AI. 4 CockroachDB tools integrated: Cloud, Vector Indexing, MCP Server, ccloud CLI.

Challenges

Designing the vector search pipeline for semantic memory retrieval. CockroachDB pgvector with cosine similarity solved this.

Accomplishments

Fully working persistent memory agent, 4/4 CockroachDB tools integrated, automatic fact extraction from conversations.

What's next

AWS Bedrock deployment, multi-modal memory, MCP direct agent integration.

Built With

• cockroachdb