Tabula blog: notes on AI memory, MCP, and owning your context

What People Actually Use Tabula For
Four reasons people connect Tabula: switching AI mid-project, keeping memory without handing it to a company, context scattered across months, and the ideas you never typed.
Read
What to Put in Your AI's Memory, and What to Leave Out
Everyone tells you to turn AI memory on. Nobody tells you what to put in it. A short guide to what earns its place, what to leave out, and why stale memory is worse than none.
Read
Your Best Ideas Don't Happen in a Chat Box
Your AI's memory is made entirely of things you typed into a chat. Capture is the way in for everything else: the book, the walk, the whiteboard. Private by default, until you decide otherwise.
Read
Add Memory to Your OpenRouter Setup
OpenRouter gives you every model. It also gives you a fresh amnesia with each one. How to run one persistent memory across LibreChat, Open WebUI, and your own scripts.
Read
How to Keep Your Context Across ChatGPT, Claude, and Every AI
A practical guide to keeping your project context, preferences, and decisions intact when you move between ChatGPT, Claude, Perplexity, Grok, Manus, and Mistral.
Read
Tell One AI. Every AI Remembers.
Tell ChatGPT you're going vegan this month. Open Claude and ask for a meal plan: it already knows. Real screenshots of one memory working across two AIs, and how to run the same test yourself.
Read
OpenAI Made ChatGPT's Memory Better. It Still Isn't Yours.
OpenAI made ChatGPT's memory much better. It still only works inside ChatGPT. The better one AI gets at remembering you, the more you need a memory you can take to the next one.
Read
Building Tabula: One Memory, Every AI
Switch AI tools and you start over. Why we are building a memory you own that moves with you across every AI, and how the open standard MCP makes it possible.
Read