Hi all - I've built @Mnexium AI and I thought the fastest way to get folks to try was it to build a chat plug-in for websites. I am providing free keys (however much usage it may be) to anyone who is willing to try it.
The plug-in can be found on NPM https://www.npmjs.com/package/@m...
We just shipped @mnexium/chat: a single npm package that adds a polished, production-ready AI chat widget to any website. React, Next.js, Express, or plain HTML it just works, and most importantly it remembers.
Most AI apps eventually hit the same wall. They forget users unless you build a ton of infrastructure first. This means every AI dev eventually will end up building this infra to provide the best user experience needs for their agent and app.
What rolling your own really means:
Vector DBs + embeddings + tuning
Extracting memories from conversations (and resolving conflicts)
Designing user profile schemas and keeping them in sync
Managing long chat history + summarization pipelines
Juggling different formats across OpenAI, Claude, etc.
Most AI memory systems treat all memories equally. Something mentioned two years ago carries the same weight as yesterday's conversation. That's not how human memory works and it creates awkward, irrelevant AI responses.
Today we launched Memory Decay, a feature that makes AI memory behave more like human memory. Frequently used memories stay strong. Unused ones naturally fade. The result is more relevant, contextual AI interactions.
When people talk about AI memory, it s usually framed from the developer s side. How do we store it? How do we retrieve it? How do we keep context alive? This is where @Mnexium AI started as well since that ecosystem is important.
But the initial vision and goal was very different and yet to be executed on.
What if users owned their memories not just the app owners?
๐ง ๐๐ง๐๐ฑ๐ข๐ฎ๐ฆ = persistent memory for LLM apps.
Add one ๐ฆ๐ง๐ฑ object and get chat history, semantic recall, and user profiles that follow users across sessions and providers.
๐ Works with ๐๐ก๐๐ญ๐๐๐ and ๐๐ฅ๐๐ฎ๐๐ โ same memories, any model. Switch mid-conversation without losing context.
โ๏ธ No vector DBs or pipelines. A/B test, fail over, and route by cost โ your memory layer stays consistent.
