LucieEveille

LucieEveille / kiwi-mem

Public

🥝 AI 伴侣记忆系统 — 向量搜索、记忆热度、Dream 睡眠整合、日历层级摘要，为 AI 陪伴场景而生。

18
4
100% credibility
Found Apr 18, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

kiwi-mem is a backend system that adds realistic long-term memory capabilities to AI chat companions, simulating human-like forgetting, consolidation during 'sleep', and hierarchical life summaries.

How It Works

1
🔍 Discover kiwi-mem

You hear about this clever way to make your AI chat buddy remember conversations like a real friend does over time.

2
📥 Get it running

Download the ready-to-go package and launch it with one easy click on your computer.

3
🔗 Link your AI helper

Connect your usual AI service so it can start building a personal memory of you right away.

4
💬 Chat as usual

Talk to your AI about your day, habits, or plans – it quietly saves the important bits as memories.

5
🧠 See the magic

Your AI starts recalling details from past chats, like your favorite foods or recent goals, feeling truly personal.

6
🌙 Let it rest and grow

Tell it to 'sleep' occasionally to tidy up memories into neat stories and future ideas.

❤️ Lifelong companion

Now your AI feels like a close friend who knows you deeply and gets better every day.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is kiwi-mem?

kiwi-mem is a Python-based proxy gateway that adds human-like long-term memory to any OpenAI-compatible LLM, turning stateless chatbots into persistent AI companions. It stores conversation fragments in PostgreSQL with pgvector, retrieves them via semantic search, and injects relevant memories into prompts—solving the problem of AIs forgetting your life details across sessions. Users get endpoints like /v1/chat/completions for seamless integration, plus an admin panel at /admin for tweaking settings without restarts.

Why is it gaining traction?

Unlike basic RAG vector stores, kiwi-mem mimics real memory with heat decay (frequent recalls boost relevance), Dream sleep consolidation (auto-merges fragments into scenes), and calendar hierarchies (daily details compress into weekly/monthly summaries). Devs hook into github dream ai features like MCP tools for memory search/save or web search across engines, plus Docker one-click deploys. The kiwi mem system cuts prompt costs via caching and tiered injection, making AI feel alive without token bloat.

Who should use this?

AI companion builders crafting personal assistants that track health habits, projects, or emotions over months. Indie devs prototyping dream ai github apps like kiwi memory game bots or memory kiwi journals. Teams needing quick persistent context for roleplay or daily journaling without rebuilding RAG from scratch.

Verdict

Grab it if you're building AI that lives with users—Docker setup and admin UI make prototyping fast, despite 18 stars and 1.0% credibility signaling early-stage maturity. Solid English README and seed memories import lower the barrier; test on a side project before production.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.