The problem: Decisions get made, but the reasoning disappears. Last month a teammate asked why we scoped a feature a certain way. I knew the answer - I'd worked through the trade-offs with Claude over 30+ messages. But that rationale lived in my head and chat history. To everyone else, the decision just appeared without a context, without "why." If I get hit by a bus, that thinking is gone.
We tried the usual fixes. Notion wiki. Obsidian. "Let's document decisions." Nobody kept it up. Too much friction. The actual reasoning stays trapped in chat threads nobody else will ever see.
What I built: Nuggetz is a browser extension that extracts decisions, insights, and action items from ChatGPT, Claude, and Gemini chats. You pick a thread, click once, and an LLM parses it into structured "golden nuggets of knowledge".
Those nuggets flow into a separate web app where your team can search them via RAG-based agent, follow specific topics, share context with colleagues, or feed them directly to other AI tools. We're building this so the switch between platforms is frictionless - your teammates and your AI tools get the right context without anyone maintaining a knowledge base manually.
Privacy note: For those wondering, the extension only processes the DOM of the specific chat you choose to extract. We don't access your other threads.
Coming soon: We're testing an MCP server internally. Once released, you'll be able to save nuggets or inject context directly from any MCP-compatible client.
Stack: Chrome extension for capture, web app (Next.js) for knowledge management, extraction pipeline using Azure OpenAI API. Early stage - works for my team, now seeing if it generalizes.
Pricing: Free for personal use. Team features will be paid eventually, but right now I want to validate whether this solves the problem for others.
What I'd love feedback on: - How do you handle the "decision got made but nobody knows why" problem? - Is decisions/insights/actions the right way to categorize what's worth keeping?