Pieces Copilot+ - Remember anything with a real-time, on-device AI assistant

@tsavo_at_pieces Are there any plans to integrate Pieces Copilot+ with other popular tools or platforms in the near future?

Congrats on the launch @tsavo_at_pieces and team! I have been using Pieces for a while and it's starting to become a mainstream tool for me. Less context switching, more productivity.

@ashikhameed you bet! We've already got a ton of integrations with dev tools that developers use every day i.e. Chrome, Visual Studio, Visual Studio Code, Obsidian, Microsoft Teams, GitHub & GitHub Gists, Obsidian, JupyterLab, RayCast, Sublime, etc.. but we're right around the corner for some amazing cloud integrations! That said, the beauty of the Workstream Pattern Engine is that is understands data streams at the OS level and doesn't need to be directly integrated to understand what's going on in each app and how it's related to the work you're doing! Try it out and let us know what you think :)

@tsavo_at_pieces Awesome AI app, guys! I like the concept you used to style it. Good luck with the product!

@tsavo_at_pieces Congratulations on the release of Pieces Copilot+ Achieve more with Pieces!📈

Hey hey @manan_gupta19 ! Most importantly it's an offline first architecture complete with 8 local LLMs 🙌 You can leverage cloud LLMs if you'd like but our goal is to give the user domain and control over their data. Take a look at some of the details here :) https://docs.pieces.app/product-... Would love thoughts!

@tsavo_at_pieces understood, that brings me some peace! All the best!

@manan_gupta19 @tsavo_at_pieces Awesome work Tsavo, was wondering how you guys make money on this ?

@manan_gupta19 @tsavo_at_pieces @archierex excellent question! We are offering it for free at the moment as we work to build an amazing product, not just for individual developers but soon for teams and enterprise. We will begin offering PiecesPro later this year and have posted a draft pricing plan here: https://github.com/pieces-app/su... Would love your feedback on this!!

@rosieatpieces Greatt! definitely will check them out

@chandula yes, it's free for now

@chandula It shouldn't have to be! Pieces for Developers is such a great product I actually want to pay for it.

@chandula yes! We're working HARD on some crazy paid features that will land end of august and early September for Teams and Enterprise! Pieces Pro will be ~$8.99/mo USD and I can promise you will be packed with some pretty awesome capabilities...

@naomigarcia Hey Naomi! I handle marketing at Pieces, and I use it everyday. Even marketers have 50 tabs open on their browser and tons of emails/messages, so it's super helpful to be able to ask Pieces anything about what I've been working on. For example, I'll ask "based on my team's conversation in Google Chat, write an outline for a blog post around x." or "write an ad based on the documentation, blog post, and web page I just referenced." As for models, we have various fine-tuned small language models that help with auto-enrichment of code when you save it to Pieces. When it comes to Pieces Copilot+, the user can choose between a handful of different local and cloud-based LLMs like GPT-4o, Llama 3, Gemini, Claude, Phi-3, etc. Pieces is completely free!

@cole_at_pieces If it's completely free, it suddenly starts sounding suspicious. As they say, if it's free, you are the product. So, are we the product? Do you not monetize it at all? Or thinking too critically, are you stealing my credit card info as I type it in somewhere?

@naomigarcia we just announced our Series A funding today! Since we're venture backed and run with a lean team of ~20 people, we still have plenty of runway to continue providing a free product. Pieces runs completely on-device, meaning you can even turn off your WiFi and use it. Plus we don't ask for a credit card anywhere, so you wouldn't be typing it in. The goal is to monetize at the right time through team and enterprise partnerships, as well as a Pro plan that Tsavo mentioned in another comment here. For now, we're just looking for user feedback and to continue working on our ambitious product roadmap so we can make the best productivity tool possible for developers!

@eddiejaoude Thanks a lot Eddie, glad you liked the launch!

@eddiejaoude thank you for all of your support!

Hey @tonyhanded, appreciate the support!

@tsavo_at_pieces @tonyhanded Thanks Tony! The next phase of Pieces will be even more proactive. We'll be leveraging Live Context to enable a real-time feed of useful developer materials and information, which will be lightning fast due to our on-device LLMs. For example, you click on a new code file, and Pieces will update the feed to provide code snippets you'll likely need to use/adapt, related URLs to help you further your research, and suggest people in your organization that are related to the problem at hand so you can pair-program. We'd love to get your thoughts on this!!

@cole_at_pieces that’s pretty darn cool!

@alex_dulub Thanks for the support!! I hope you love it when you get a chance to give it a test drive. We ensure data privacy by being local first. Our entire architecture is centered around being able to use Pieces completely offline. When it comes to Live Context, everything is captured, processed, and stored completely on-device. If you use a local LLM like Mistral as your Pieces Copilot's runtime, none of that context will ever leave your device! If you do choose to use a cloud model, know that only the very specific context related to your prompt will be leveraged. We have a really excellent blog that describes this in detail here: https://code.pieces.app/blog/an-...

@tsavo_at_pieces @chrissycodes Thank you @chrissycodes!! We really appreciate the support you've given us!!

@tsavo_at_pieces @chrissycodes Chrissy, your articles have been really great, thanks for always supporting Pieces!! 🥹

@francesco_ciulla lets goo Francesco, so happy to be alongside you today 💛

@francesco_ciulla thank you so much for all of your support!!

@rosieatpieces let's go!

@keimaumau thank you SO MUCH for being an early adopter and for sticking with us this long!! That is a good point - like you mentioned, Pieces requires more resources than it used to due to the fact that we are running everything on-device for privacy and security purposes. You can disable the Workstream Pattern Engine (the magic behind Live Context) at any time to reduce the load. However, the release we sent out last night should have improved this a decent amount, so please try it out today and let me know what you think!

@andreas_sohns we are so glad you think so, being local first is so integral to our company and our goals!

@daniel_d7 essentially yes it can, but only what you allow it to see and entirely on-device! If you enable the Workstream Pattern Engine, it will be able to extract key information using a vision processing stack, but only from what you have on your screen. All of this is captured, processed, and stored entirely on-device, and then the Pieces Copilot can leverage this information to give you seriously helpful responses! For more details, check out this article: https://code.pieces.app/blog/an-...

@rosieatpieces It's a really nice idea! Isn't Microsoft is doing something similar for Windows 11 ?

@daniel_d7 Yes, some other companies are hoping to achieve a similar goal in being able to extract segmentations of the foreground application, however they are not leveraging this in the highly contextual and workflow-specific way we are, and they are most certainly not doing it entirely on-device like we do!

@rosieatpieces Ok thanks for the explanation. Is there any price for that? I can't find any mention of that

@daniel_d7 We are offering Pieces for free at the moment as we work to build an amazing product, not just for individual developers but soon for teams and enterprise. We will begin offering PiecesPro later this year and have posted a draft pricing plan here: https://github.com/pieces-app/su... Would love your feedback on this!!

@tsavo_at_pieces @ashikhameed Thanks a lot for your feedback! Do let us know if you have any questions when trying out Pieces

Thanks for the support! @kehui_guo Looking forward to see your use-cases across your tool chain :)

@kehui_guo we are trying our best to be a truly horizontal and completely integrated workflow assistant! Thank you for the kind words!

@tsavo_at_pieces @andrewbaisden thank you so much for the support!!

@memphys_sk Pieces Copilot+ with Live Context ensures your data remains secure and private by leveraging on-device processing and intelligent data handling techniques. This approach provides a secure, efficient, and privacy-focused tool for enhancing developer productivity. Here are the main features: Privacy and Security at the Core:On-Device Processing: All workflow data is processed and stored locally on your device, ensuring sensitive information never leaves your machine unless explicitly shared. Intelligent Visual Snapshots: Instead of continuous screen recording, PiecesOS captures intelligently timed snapshots of application visuals, processed on-device to ensure only relevant information is analyzed. Temporary Data Handling: Extracted text is temporarily stored in memory and deleted after processing, typically within 100 milliseconds. Irrelevant data is discarded after 12 hours. Data Compression and Redaction: Relevant data is compressed to about 10% of its original size using an on-device transformer model, filtering out sensitive information like API keys and PII. Secure Local Storage: Summarized data is stored in a local vector database on your device, queried during retrieval-augmented generation (RAG) sessions without transmitting data to the cloud. Optional Cloud Integration: You can use cloud-based LLMs if preferred, but only refined context is sent to the cloud. On-device LLM runtimes ensure maximum privacy. Secure Integrations: For integrations like the VS Code Plugin, data exchanges remain secure via local HTTP/GRPC connections.

@meni_teknikio Thanks for your support! Let us know what you think about the product!

Hey there @kyrylosilin Key here is that our architecture is offline first. You can choose from 8+ local LLMs (my favorite right now is LLAMA 3) to handle all of the processing locally. You can leverage cloud LLMs if you'd like but our objective is to give users domain and control over their data. You can check out this page for more information on privacy and security: https://docs.pieces.app/product-... As for use cases, the sky is the limit. Personally I love: - generating github issues from emails and chat - fixing and iterating on code without having to copy paste what i want to work on into chat - summarizing research papers and finding common themes in a body of research Would love to hear more of your thoughts as you experiment!

@sinikiwe_phahlane Thanks a lot for your response. Glad you find it useful!