Super simple to use through source or the binary. Had fun playing at lunch on the mac mini. looking forward to putting this on my gamin' computer with the 4090.
Great interface and made it super simple to download models. Fun project
Hey Product Hunt community! 👋 I'm John, and I'm thrilled to be here meeting everyone.
I'm excited to introduce you to Klee, a secure AI assistant that runs locally on your desktop.
We believe AI is a revolutionary productivity tool, potentially more impactful than operating systems. Many companies want to embrace AI, but data security concerns are holding them back. Their data regulations often prohibit them from sending sensitive information to cloud APIs.
That's why we developed Klee. It offers on-device processing, providing comprehensive insights while ensuring complete data security and privacy. Key features include:
💕 Local Data Processing - The RAG (Retrieval-Augmented Generation) process happens entirely on your device. No data is uploaded to or downloaded from the internet.
🚀 Native macOS App - We've built Klee using Swift and SwiftUI, ensuring high performance and simplicity.
📑 Open Source LLMs - We've integrated powerful open-source Large Language Models, such as Llama 3.1.
We created Klee because we believe privacy is crucial in the AI age.
I'm here to answer any questions and would love to hear your feedback, either in this thread or on our X (formerly Twitter) account: https://x.com/klee_ai_desktop.
Thank you for your support and for being such a fantastic community!
Report
@john_weizhong Hey John, congrats on launching Klee! A secure, on-device AI assistant sounds very promising. Here are some questions and how I can help:
- How does Klee ensure data security on the user's device? Are there any encryption methods implemented to protect sensitive information?
- How are Klee's AI functionalities currently trained? Is there potential for user-specific training or customization of its responses?
As a developer, I'm interested in the security architecture behind Klee and the potential for further AI training features. If those align with my skills, I might be able to contribute to future development.
Overall, Klee seems like a valuable tool for privacy-conscious users who want to leverage AI assistance. Exploring robust security measures, user-specific training options, and potentially expanding its functionalities could further enhance its value proposition. I'd be happy to see how Klee evolves and potentially contribute my skills if relevant!
Super cool on device model! LLM on edge is here to stay. Curious how’s your performance compared to octopus-v2?
Report
Hi hi hi... How can I please get more documentation on how to use it? I've added some content (folders from my machine) but not sure how to use it properly. I've asked it questions related to the files I've shared with it - but it isn't understanding (or I'm not).
@yomiko Thanks, it took a long time for the team to make Klee hope you will enjoy it.
Report
Hey congrats @john_weizhong and launching! Gotta love on-device AI! With the release of OS LLMs like Llama 3.1, honestly, on-device could be faster and gets similar results to at least GPT 4 level responses. Great idea to focus on privacy!
@tonyhanded Thanks Tony, exactly, open source LLMs are more than capable of manging local knowledge, and privacy is ensured cause no data is collected.
Hi hi hi... How can I please get more documentation on how to use it? I've added some content (folders from my machine) but not sure how to use it properly. I've asked it questions related to the files I've shared with it - but it isn't understanding (or I'm not).
Thank you so much!
-Paul