Skip to content

Memory gap

Bora Celik
Bora Celik
2 min read
Memory gap

Folks I deeply respect, like Scott Belsky, say memory will be the ultimate moat in AI.

We have seen ChatGPT and Claude release their memory features so they remember things about us. This makes us want to use their product more because with personalization the results are better.

But our personal memories tied to these AI assistants creates vendor lock-in and is bad for us as consumers. I should be able to keep my memories in a neutral place and take it with me to any AI assistant/agent that I use.

So that I don't have train everyone's AI apps with my personal data every time. I own my data and should be able to control it.

This is why I built Memkit as a portable long-term memory MCP server that I can bring with me to any AI assistants that I use.

Here is an example of how I use it in Claude.

and in a different chat:

For books:

and then:

It saves memories as vectors in my Memkit account and searches them when the AI assistant needs them. It's important to surface relevant memories for the occasion instead of all memories or only recent memories, so vector search is good for this.

It works pretty good with my voice and Whatsapp AI assistants as well.

DM me on Linkedin if you'd like to try Memkit.

Bora

Article

Bora Celik

Lifelong student of bringing original ideas to life. Surf nut. Founder @ Collabs.io / Subkit.


Related Posts

Members Public

gull

Did I just build world's simplest calories tracker or no? Try gull.fyi and let me know. 0:00 /0:33 1× I couldn't stay disciplined at meal tracking using OG apps so made this. Just send text/voice/photos to WhatsApp +1.401.566.7960

Members Public

crow

Finally solved my new book discovery problem by building an AI agent. Glorious results! Try it and tell me if it saves you a ton of time too: crow.fyi Here is why I built this and how it works. Every Tuesday, new books flood the market. As a voracious

crow
Members Public

Idea check

$22M+ raised, 100+ VC pitches, 12 years as startup founder. I've learned a few things about ideas that can get big. Is your idea worth building in 2025? The answer has changed. I built an AI idea evaluator to help: ideacheck.dev My scoring framework (each criteria: 0-2

Idea check