Kortex - Organize, import + export NotebookLM sources with a click
by•
Most extensions are basic web clippers. Kortex acts as an integration layer, bridging active AI chats with NotebookLM storage. Its standout features include Native LLM Archiving, which adds a one-click import button directly inside ChatGPT, Claude, and Gemini. Unlike others, we focus on data freedom with bulk exports to Markdown, and precision sniping to save specific text blocks rather than whole messy webpages.
Replies
Kortex
Triforce Todos
Any plans to support custom destinations beyond NotebookLM?
Kortex
@abod_rehman Do you have anything in mind? It might me a good idea to expand beyond NBLM.
Congrats on the launch! The 'Chat with PDFs' feature is crucial for a second brain app.
I’m curious, how are you handling the context retrieval for larger documents? Do you have a way to prevent the AI from hallucinating details not found in the source text? I’m tackling this exact challenge right now with my own tool (SquarePact) for legal contract review, so I'm always interested in how other makers ensure high-fidelity retrieval.
Excited to try out the new workflows!
Kortex
@anighojkar I think you misunderstood my product's goal. It does not produce AI responses, it helps to take LLMs like claude,chatGPT conversations into the notebookLM.
Clean concept. But what’s the real edge here — is it the native LLM archiving or the precision-sniping workflow? Also, how do you handle formatting consistency when clipping AI chats, and what stops this from becoming just another fancy web clipper? Curious how Kortex stays genuinely differentiated long-term
Kortex
@adityaraj007 The real edge is the workflow: We treat the capture (clipping) as just step one. The differentiation is how seamlessly that captured content flows into your writing and thinking environment without context switching.
regarding formatting: We parse the HTML specifically to convert it into clean [Markdown/Blocks] so it remains editable and structured in your notes. Long term, Kortex isn't about better bookmarking; it’s about reducing the friction between consuming information and creating new work.
Swytchcode
Quite amazing. Do you also preserve the text formatting of the chat generated by LLMs?
Kortex
@chilarai Yes! Each chat is formatted as user question and model response.
Smart move. NotebookLM is great but feels incomplete without easy export. Building on top is the play.
Question - does it work with other AI chat sources too? Or just NotebookLM for now? Also, any plans for a web version instead of just chrome extension?
Clean product! 🙋
NotebookLM is powerful but there are still those tiny good-to-have features that we usually miss.... and Kortex seems to be bridging that gap... Interesting...
Best luck on the launch!
Kortex
@dkspeaks Thanks, Dilip! That gap is exactly where we live. We want to take the 'friction' out of using these powerful AI tools so you can just focus on the work. Appreciate the kind words!
How does export work? Can I bulk export NotebookLM sources to Markdown/JSON and keep tags/folders, and do you handle de‑duplication if I import the same ChatGPT/Claude thread more than once?
Kortex
@marryann Hi Mari! Great questions. We believe you should own your data, so:
Export: Yes, you can bulk export sources to Markdown/JSON. We preserve the folder structure wherever the API allows us to detect it.
De-duplication: If you import the same ChatGPT thread twice, we detect the unique thread ID. Instead of creating a copy, we update the existing file with any new messages.
Are there plans to expand beyond Chrome to support other browsers like Firefox or Edge？
Kortex
@lightninglx Thanks, Xiang! Since Edge is built on Chromium, you can actually install the current extension there right now—it works great!.
As for Firefox, it requires a different architecture, so it is on our roadmap for later this year. We want to nail the experience on Chromium browsers first before branching out.
Wow, Kortex looks amazing! The NotebookLM integration with one-click archiving from ChatGPT is seriously clever. How does it handle overlapping or conflicting sources from multiple chats?
Kortex
@jaydev13 Do you mean sources from ChatGPT?