dahsmartgirl/oml-mvp

OML – Open Memory Layer v0.1.0

One memory layer. Every AI chatbot.

The Problem

You tell ChatGPT about your project.
Then you switch to Claude and explain it again.
Then Gemini. Then back to ChatGPT in a new session.

Every AI treats you like a stranger.

Your conversations are siloed. Your context is scattered. You explain yourself over and over.

The Solution

OML creates one unified memory layer that works across all AI chatbots.

Save context once. Use it everywhere.

Tell ChatGPT about your startup → OML saves it
Switch to Claude → OML injects that same context
New ChatGPT session tomorrow → OML remembers

Your memories follow you. Not trapped in one app.

What It Does

Unified Memory Across All LLMs

Save memories once, use them in ChatGPT, Claude, Gemini, or any future AI
No more re-explaining yourself across different chatbots
Your context is portable, not siloed

Own Your Data

Everything stored locally in YOUR browser
Export anytime (coming soon)
No company owns your conversation history
Future: Sync to YOUR cloud (Google Drive, Dropbox, iCloud)

Visualize Your Memory

See all your saved context in one place
Search, filter, and organize by tags
Understand what you've told AIs over time
Track your conversations across platforms

Smart Organization

Auto-tags your memories (project, code, personal, etc.)
Quick-edit tags and summaries
Filter by topic or source

Privacy First

100% local storage – nothing sent to servers
Open source – verify the code yourself
You control what gets saved

Built for the Future

APIs and SDKs coming
Third-party integrations planned
Extensible architecture and many more.

Import Existing Chat History

Got old ChatGPT or Claude exports?

Click OML icon
Click "Upload"
Select your .json export file
OML extracts facts automatically
Now accessible across all AI platforms

Install (2 minutes)

Chrome/Edge/Brave

Download this repo
- Click green Code button → Download ZIP
- Unzip anywhere
Load into browser
- Chrome: chrome://extensions
- Enable "Developer mode" (top-right)
- Click "Load unpacked"
- Select the unzipped folder
Done!
- See the icon in your toolbar
- Go to ChatGPT or Claude to try it

How to Use

Save Memory (Works Anywhere)

Right-click method:

Highlight text on any webpage
Right-click → "Save selection to OML"
Auto-tagged and saved

Quick-edit method:

Highlight text
Click "Save" button that appears
Shift+Click to add custom tags

Manual entry:

Click the icon in toolbar
Add text and tags
Save

Use Across Different AIs

Scenario: You told ChatGPT about your project

Switch to Claude.ai
OML banner appears automatically
Click "Insert memory"
Your project context from ChatGPT is now in Claude

No re-explaining. Your memory follows you.

Manage Everything

Click the icon to:

View all memories from all AI chats
Search by keyword or tag
See what you've told ChatGPT vs Claude
Edit, organize, or delete
Track your conversations over time

Why This Matters

Right Now:

You → ChatGPT: "I'm building a startup..."
[ChatGPT remembers in its silo]

You → Claude: "I'm building a startup..." 
[Explain again, Claude has separate memory]

You → New ChatGPT session: "I'm building a startup..."
[Explain AGAIN, memory was lost]

With OML:

You → ChatGPT: "I'm building a startup..."
[OML saves it]

You → Claude: [OML auto-suggests that context]
[Click insert, Claude now knows]

You → New ChatGPT session: [OML remembers]
[Your context persists across sessions]

Your memory becomes platform-agnostic.

Supported AI Platforms

Currently works on:

✅ ChatGPT (chat.openai.com, chatgpt.com)
✅ Claude (claude.ai)
✅ Gemini (gemini.google.com)
✅ Grok (grok.com)

Save from anywhere, insert into any of these.

More platforms coming soon.

Roadmap

Phase 1 (Now) – Foundation

Phase 2 (Next) – Your Data, Your Cloud

Export all memories (JSON, CSV)
Cloud sync (Google Drive, Dropbox, iCloud)
Memory visualization dashboard
Better relevance matching
Bulk edit and organize

Phase 3 (Future) – Open Platform

Vision – Universal Memory Layer

Works with ALL AI chatbots automatically
Developer platform for building on OML
Memory marketplace (share templates, prompts)
Multi-modal memory (voice, images)

Data Ownership

Your data belongs to YOU

Stored locally in your browser
No servers, no company databases
Export anytime
Future: Sync to YOUR cloud accounts

Not like other "AI memory" tools

Other Tools	OML
Data stored on their servers	Local-first, your device
Works with one AI only	Works across all AIs
Closed source	Open source
Subscription required	Free forever
You can't export	You own everything

Tech Stack

OML v0.1.0 is built with a simple and robust stack, with no external frameworks.

Core Logic: Vanilla JavaScript (ES6+)
UI: HTML5 & CSS3
Storage: chrome.storage.local API for 100% local-first data.
Packaging: Manifest V3 for a modern and secure Chrome Extension.

For Developers

Coming Soon: APIs & SDKs

We're building OML as an open platform:

// Future API (not implemented yet)
import { OML } from 'oml-sdk';

// Save memory programmatically
await OML.save({
  text: "User prefers React over Vue",
  tags: ["code", "preferences"],
  source: "my-app"
});

// Query memories
const relevant = await OML.query("react framework");

// Inject into any LLM
await OML.inject(relevant, { target: "chatgpt" });

Want to contribute? Open an issue or PR.

FAQ

Q: Why not just use ChatGPT's memory feature?
A: ChatGPT memory only works in ChatGPT. OML works everywhere. Plus you own the data.

Q: Where is my data stored?
A: Locally in chrome.storage.local. It never leaves your device unless YOU choose to sync.

Q: Can I use memories from Claude in ChatGPT?
A: YES. That's the whole point. One memory layer, every AI.

Q: Is this like a second brain / PKM tool?
A: Similar idea, but specifically built for AI context. Think Notion/Obsidian but for LLM memory.

Q: Will there be a mobile version?
A: Yes, mobile keyboard extension is planned (Phase 2).

Q: Can I self-host the API?
A: When we release the API, yes. It's open source.

Q: How is this different from RAG?
A: RAG is for documents. OML is for personal context and conversation memory.

Known Limitations

Memory relevance uses keyword matching (can be improved)
Desktop only (mobile coming)
Manual insertion recommended (auto-suggestion being refined)

This is v0.1 – early but functional.

Contributing

This is open source. PRs welcome.

Want to help?

Test it and report bugs
Suggest features (open an issue)
Contribute code (fork + PR)
Share it with others

License

MIT License – Use it however you want.

See LICENSE file for full terms.

Why I Built This

I was tired of AI cross amnesia.

I'd spend 10 minutes explaining my project to ChatGPT.
Then switch to Claude and do it again.
Then start a new ChatGPT chat and explain AGAIN.

My context was scattered across platforms, trapped in silos.

I wanted ONE memory layer that worked everywhere.
I wanted to OWN my conversation data.
I wanted to see all my AI interactions in one place.

I am not a dev but i had to do something.

So I built OML.
This is the beginning of portable, user-owned AI memory.

Star this repo ⭐ if you believe AI memory should belong to users, not platforms

Built with frustration and determination by Ilerioluwa🌟

Questions? Open an issue.