Pieces Long-Term Memory Agent - The first AI that remembers everything you work on

@chris_lane_jones Thank you for the support Chris! That is a great way of using it, searching through old chats and email threads can get really tiring.

After you try LTM-2, we would love to hear what you think about it 🤝

@slavisa_dujkovic Even non-developers can use Pieces to help with retaining all of their workflow context! 😉 If you end up giving it a try, we'd love to hear what you think!

@soumikmahato thanks, yeah definitely an upgrade over existing AI agents.

@soumikmahato Thank you Soumik! Definitely better than the traditional AI, if you end up trying Pieces, we would love your feedback! 🙏

Thanks@sebastian_voletti, your support means a lot;

@zijian Thank you for your support Gong 🙌

@zijian Thank you for the support! Good luck on your upcoming launch 🙏

@emmanuelisenah Thank you for the support! And yes! With LTM-2 you can even ask about last month, or about a project you were working on 3 months ago!

@joewardpr Wow! After trying Pieces we would love to know if it lives up to what you've been thinking about all these years. Also, if you have any suggestions for how we can improve it we're all ears.

Thanks so much for the support!

@elliezub I haven't tried it yet, but I recommend that you do a rapid-fire video that showcases 5-10 things you can do with Pieces very quickly, no intro dialogue or bloat.

The part that jumped out was "remember everything you work on." I'm not sure if that is user-driver (actively selecting pieces to store) or if it observes activity and knows what to record automagically.

I worked on a prototype Chrome extension that would allow me to save links and snippets of a page, and assign it to a "project" while I coded, so that I could have a reference to the research I was using during development. That is never captured in GitHub (just the result of all that research put to work). But, it still has value. At the time I did the prototype, there were no LLMs like we have today to enhance the capabilities, and it resulted in being a bit of a mixed-format bookmarking/note-taking extension. It was monitoring my activity.

Your plugin model is essential. I imagined (minimally) that something (a plugin) was monitoring Terminal, my browser, and my IDE. And it automatically recorded my activity, knowing what was relevant to my key task, and even associating some of the research done to specific segments of the code. I also imagined that it could auto-summarize the research in a file within a coding project structure, that would get included in the GitHub repository has it changed. And that would basically tell the story of historical changes and development to the project, but more about process and research than just a change log.

I don't think you've achieved this all yet, but you're probably very close. At the very least, you're doing multi-plugin, user-requested storage of resources (pieces), with LLM summarization and questioning.

Hi@martin_streicher1 ,
We have no plans to add Audio to Pieces in the near future, however, if you turn on closed captions in meetings or on videos, Pieces does a great job at keeping up with the conversation!

@jakejohnson21 Thanks for checking out the launch Jake!!

As for privacy, Pieces keeps everything local by default, so your data stays on your device. Sensitive details get filtered out automatically, and if you are using a local LLM like Mistral or Phi, nothing ever leaves your machine. If you go with a cloud model, Pieces only sends the context needed to answer your question.

Hope that helps! If you're interested in learning more about privacy, you can check out this page in our docs: https://beta.docs.pieces.app/products/privacy-security-your-data

@jacob_molz1 Hi Jacob! Thank you for supporting our launch!

I know you used LTM a few times in the past, so I know the team would absolutely love to hear what you think of LTM-2 considering all of the updates that have gone out since our last chat. If you ever want another intro to Pieces call, feel free to reach out! 🙏