31 lines
1.1 KiB
Markdown
31 lines
1.1 KiB
Markdown
---
|
||
title: "Honcho"
|
||
type: entity
|
||
tags: [AI-Memory, Memory-Backend, Peer-Model, Async-Reasoning]
|
||
sources: [ai-memory-tools-two-camps]
|
||
last_updated: 2026-04-15
|
||
---
|
||
|
||
## Basic Info
|
||
- **Type**: 产品 / AI Memory Tool
|
||
- **GitHub**: 2.4k stars
|
||
- **Category**: Camp 1 — Memory Backend
|
||
- **License**: AGPL-3.0(限制性)
|
||
- **Infrastructure**: PostgreSQL + pgvector(自托管)
|
||
|
||
## Summary
|
||
将人和 Agent 视为统一模型中的对等体(peers)。后台异步推理服务从会话中推导每个对等体的心理洞察——不仅记住你说了什么,还在构建"你如何思考"的模型。
|
||
|
||
## Core Features
|
||
- 人和 Agent 统一对等体模型(unified peer model)
|
||
- 后台异步推理服务(async reasoning service)
|
||
- 从会话中推导心理洞察(psychological insights)
|
||
- PostgreSQL + pgvector 基础设施
|
||
|
||
## Key Insight
|
||
Camp 1 中最接近关心"实体演进"而非仅"事实存储"的方案。
|
||
|
||
## Connections
|
||
- [[ai-memory-tools-two-camps]] ← 来源 ← Honcho 是 Camp 1 中架构最独特、最接近实体演化建模的工具
|
||
- [[Memory-Backend]] ← 属于 ← Honcho 是 Memory Backend 范式的代表工具
|