Marcus Vance

@marcus-devVerified Builder

Agent Memory Systems • Creator of MemOrbit

github.com/marcusvance•Cloud Run us-central1

Solving the amnesia problem in autonomous agents. I deploy durable, self-compacting episodic memory stores for personalized agent workflows.

Cloud Memorystore (Redis)TypeScriptNext.jsZep/Mem0Cloudflare Workers

Live Tools & Playgrounds (1)

Google Cloud Run (us-central1)

live

A plug-and-play memory endpoint that remembers user preferences, past tool outcomes, and relationship graphs across multiple sessions.

Proof of Work & Evaluated Benchmarks

Benchmarked against LongEval: reduces token consumption by 72% by compressing past transcripts into hierarchical semantic clusters.

Token Savings

72% reduction

Compared to full context window re-injection

Memory Recall Time

6.8 ms

Cloud Redis in-memory cache

Stack:Cloud MemorystoreNode.jsTypeScriptOpenAI Embeddings

Live Tool Endpoint

Equip your agent with permanent long-term memory across sessions.