---
title: "Discussion on \"Three Memory Architectures for AI Companions\" | Hashnode"
description: "Discussion on \"Three Memory Architectures for AI Companions\". How we built three memory systems for AI companions. pgvector with importance scoring, an LLM-managed scratchpad, and a filesystem agents navigate with bash."
url: "https://hashnode.com/posts/three-memory-architectures-for-ai-companions/69d0834c9d446ec5bd88855c"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Three Memory Architectures for AI Companions](https://emotionmachine.hashnode.dev/three-memory-architectures-for-ai-companions?utm_source=hashnode&utm_medium=feed)

How we built three memory systems for AI companions. pgvector with importance scoring, an LLM-managed scratchpad, and a filesystem agents navigate with bash.

Apr 4 · 9 min read · emotionmachine.hashnode.dev  
Read the article

-   [#ai](https://hashnode.com/tag/ai)
-   [#llm](https://hashnode.com/tag/llm)
-   [#machine-learning](https://hashnode.com/tag/machine-learning)

## Written by

[Emre Sarbak](https://hashnode.com/@emotionmachine)

@emotionmachine · San Francisco, USA

  
[View profile](https://hashnode.com/@emotionmachine)

Building conversational AI at Emotion Machine. Voice models, AI characters, agent plugins.

Co-founder at Emotion Machine, a conversational AI company working on how AI converses and builds relationships. We build Personality Machine (AI character infrastructure), domain-specific speech-to-speech voice models, and agent communication tools like Claw Messenger.

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
