Pieces Long-Term Memory Agent - The first AI that remembers everything you work on
by•
Ever wish you had an AI tool that remembered what you worked on, with who, and when across your entire desktop? Pieces Long-Term Memory Agent captures, preserves, and resurfaces historical workflow details, so you can pick up where you left off.
Replies
Fable Wizard
This is awesome! I’m always wasting time trying to find old notes, code, or links, so having an AI that actually remembers everything sounds like a huge help. How does LTM-2 keep my data private when it’s tracking work across different apps?
Pieces for Developers
@jonurbonas Hi Jonas! Pieces definitely is a huge time-saver 🙏
Also, great question! We ensure data privacy with a local-first approach, so everything Pieces Long-Term Memory captures is processed and stored entirely on your device. Sensitive details are filtered out automatically, and if you use a local LLM like Mistral, your data never leaves your machine. Plus, if you choose a cloud model, Pieces only sends the specific context needed to answer your question! If you're interested in learning more, check out this page on our docs: https://beta.docs.pieces.app/products/privacy-security-your-data
Great app, highly recommend!
Pieces for Developers
@angelos_katrantzis Thank you for the support Angelos! 🤝
@elliezub My pleasure, keep up the good work!
Pieces for Developers
@angelos_katrantzis Thank you Angelos! So excited to see how LTM fits into your workflow!
@jackross Thanks, me too!
I really took my time to write a comment about the LTM-2 because I wanted to thoroughly test its features, and I have to say, this version is absolutely incredible. The fact that it can store my workflow history for up to nine months is mind-blowing, and even better, all the data is stored locally on my system.
Another aspect worth highlighting is the addition of new control options like 'Sources' and 'Time Range,' which weren't available in LTM-I. These features make a significant difference in how I use the tool compared to before. Speaking of performance, the speed improvement is remarkable. With LTM-I, there were instances when retrieving the most recent data failed, or when it worked, it often took an excessive amount of time to generate a response. However, those issues seem to be completely resolved in LTM-2.
On a side note, I've been reading articles and blogs about Pieces, and I'm genuinely excited to try out LTM-2.5 and LTM-3 in the near future!
PS: I tried posting this yesterday, but it didn't go through. Today, I asked Pieces, and he wrote everything exactly as I had written it before.
Pieces for Developers
@grenish_rai Really appreciate your thoughtful reflection comparing LTM-1 and LTM-2. We're continually working to enhance the experience, making it even more magical and, importantly, extending it to teams and formalizing the product (and brand) beyond just developers. Small side note there — LTM-2 was actually designed from the start to benefit everyone and technically digital workers at large can use it today and leverage the power of artificial memory in their own unique workflows.
Also—sharing some insider information on what's coming up next with LTM-2.5 and LTM-3 below 👇 😉
Looking ahead, LTM-2.5 will tap even deeper into OS-level data, providing significant enhancements in free-form retrieval, temporal awareness, and effortless sharing and retrieval of memories across individuals and teams. You'll experience intuitive, dynamic summary generation based not on fixed intervals, but tailored around topics, tags, specific time ranges, and more. Additionally, navigating and browsing these workstream summaries will be substantially more intuitive and user-friendly. We're roughly six weeks out from the official release of LTM-2.5 and can't wait to hear your thoughts!
LTM-3 will push the boundaries even further, focusing on extremely deep recall capabilities—perfect when the exact timing or source of context is unknown. You'll be able to ask compound questions like, "Give me all the errors I've encountered over the last several months," or "Show me all the NPM packages I researched or used while working with Mark on the Y Project." Essentially, LTM-3 will support these complex queries and similarity retrieval across dynamic, non-deterministic timeframes.
Anyway, know that we're hustlin' & stay tuned 🙌
Appreciate you!
-Tsavo
Pieces for Developers
@grenish_rai This is awesome to hear! Really appreciate you taking the time to test LTM-2 so thoroughly. Storing up to nine months of workflow history (all locally!) was a huge focus for us, so it's great to see that resonating.
Also, thank you for all the support and being such an amazing member of our community! 🙏
Very interested to try this I am always hitting memory limits on chatgpt so this will help a lot
Pieces for Developers
@bayanmas2 Definitely give Pieces a try! We would love to hear what you think of it 😄
Pieces for Developers
@bayanmas2 Thanks for the support Bayan! Pieces gives you access to 9+ months of OS-level context. This is REAL memory!
How does Pieces utilize contextual understanding to generate ready-to-use code and explain entire workspace effectively?
Pieces for Developers
@chukwunonso_prosper Good question! Basically Pieces uses LTM-2 and Workstream Activity to save up to 9 months of your workflow history, capturing code, references, and decisions automatically. It understands your past work, generates code based on all of the saved context, and provides detailed explanations by linking back to relevant discussions. Plus, Workstream Activity gives a timeline view of recent tasks, making it easy to track changes, revisit decisions, and find important links.
Hope this helps! And if you decide to try Pieces, we would love to get your feedback 😄
sounds really interesting, especially the recall part. but how detailed is it? like if I was troubleshooting something and googling fixes, would it remember those searches along with my code changes? that could be really useful
Pieces for Developers
@pkpraveen603 Hey there Praveen thanks for the support! This is actually one of my favorite things to use Pieces for. And the best part about it is that you don't need to take time explaining everything - Pieces just knows. You can ask questions about the docs you're reading online, the code you're actively editing, and even emails about the problem - and Pieces will give you back real time summaries, links, and ways of fixing your code that pulls info from all the sources I mentioned! (Even ways you could've overlooked!!)
Our new Workstream Activity feature is super helpful for this exact kind of work too!
Pieces for Developers
Here are some of my favorite prompts to use with Long-Term Memory:
What was I working on yesterday?
What didn't I do yesterday that I was supposed to?
Who was talking about X in that meeting?
Where did they tell me the file was located?
When was the last time I saw this error and how did I fix it?"
Feels futuristic asking AI these questions but at the same time like it was the obvious missing piece in an AI assistant that actually knows you and grows with you. I'd love to hear some prompts you're using with LTM!
Pieces for Developers
@jackross Love these prompts!! See some of my favorites in there as well 🤩
Pieces for Developers
The smartest person in the room isn’t the one who knows the most. It’s the one who remembers what’s relevant at the right time. Pieces LTM 2 helps you become that person.
Pieces for Developers
Absolutely @thesmitpatel;
Looks like such a great tool for version control too!
Pieces for Developers
@aude_gleef It's saved me quite a few times after deleting code that I definitely should've kept! "I definitely won't hit that error again" famous last words!
Congrats guys, I’ve been waiting for this for a while and glad to see it finally come. Tired of having to re-explain things to ChatGPT as it is so forgetful, unless you specifically tell it to remember.
Pieces for Developers
@simonprutton Appreciate it, Simon! Yeah, constantly re-explaining things is a such a time-waster. With LTM-2, you don’t have to worry about that, your work, notes, and conversations are always there when you need them. Hope you get a chance to try it out! We'd love to know what you think about it 😄
Pieces for Developers
@simonprutton Thanks for the support Simon! Pieces remembers and you don't even have to tell it to!