There is a lot of buzz around AI tools in the development world, but very few tools solve the real issues developers face while being functional and practical products.
One such tool is Pieces (Yes, it's called Pieces, which sounds like some puzzle or game, right?). I will talk about it in this blog and share a different perspective on how it helped me from coding to content creation (no, not generating content with the help of AI).
So, What is Pieces?
Pieces is an AI-powered on-device productivity tool that serves as a companion and a code manager. It offers various functionalities such as global search, Live Context, saving code snippets, and more. It helps you be more productive and integrates smoothly with your development environment. Most importantly, it helps overcome the issue of continuous context switching, which we developers don't like.
What I Use Pieces For
I've been using Pieces for a while for multiple purposes, from saving snippets to summarizing the work I've done throughout my day. I will talk in more detail about each use case going forward.
I am a big fan of the native Pieces Desktop App for Mac, and there are also extensions and plugins for Chrome, VS Code, and Obsidian. You can find all the plugins here. Choose whichever you prefer and are comfortable with. Let's explore these features in more detail.
Saved Materials
As a DevOps professional, there are always some configurations or code snippets we often use in our day-to-day development workflow. The saving snippets feature is super helpful. Not only does it save snippets like a normal GitHub Gist, but it also automatically adds labels, and supporting articles, and creates shareable links. I can also open a Copilot Chat to ask for improvements or explanations with examples.
My favourite feature is that it summarizes what the snippet does as soon as you paste it with the help of AI. We often forget why we saved a piece of code.
Talking about the content creation I mentioned at the start of the blog, whenever I find some interesting code, I often share code snippets on social media. It's easy to lose the code snippets or save them. Now, with Pieces, I just add a label with content so that it helps me filter the snippets I want to share.
Live Context
With Live Context, Pieces gains the ability to look at the things I was working on before. It can help you pick up right where you left off, fixing a bug in a project you're developing locally or continuing work on a previous task. It's incredibly handy because you don't need to copy and paste all the files and folders or explain everything; it automatically understands everything.
One of my favourite ways to use it is to summarize the things I worked on recently. For me, this is sometimes really hard to do manually.
Running LLM On-device
We've all encountered the chaos when ChatGPT is down, right? Local LLMs can be a game-changer in those situations. However, running local LLMs can sometimes be tricky, especially for beginners.
This is where the powerful feature of Pieces comes into play—it lets you choose the Copilot Model and runtime. In simple words, you can choose whether you want answers from cloud LLMs like ChatGPT, Gemini, etc., or use on-device models such as Llama 2, Misral, etc.
This is not the end of it; they have more great functionality like Global Search, Workflow Activity, etc.
Conclusion
I am impressed by the folks at Pieces for bringing this product together and how simple it is to use. At the same time, it tackles the key areas we developers usually face, providing solutions all under one roof. This is especially valuable since we developers don't like context-switching.
I see myself using Pieces to tackle my daily tasks and I can't recommend it enough—it's definitely worth trying out.
I'll end this blog by providing some feedback and suggestions on the things I like and areas that can be improved.
Things That Can Be Improved
- The intro screen can be more simplified so that beginners don't find it overwhelming when they first try it out.
- Syntax highlighting in saved snippets for Dockerfile and other file types can be enhanced.
- Downloading snippets as images would be useful so that when I'm sharing code on social media, I don't need to switch to another tool for this purpose.
Things I Like
- Privacy through on-device processing.
- How simple it is to use and switch between different functionalities.
- The Live Context feature I mentioned above.
- The ability to switch between different models like ChatGPT, PaLM 2, Gemini, etc. This helps prevent vendor lock-in (I hope that’s the right term).
- Support for LLM locally, so you don't need to worry if ChatGPT is down.
- Last but not least, a shoutout to their team. I've met a lot of founders and tools, but the people at Pieces are really passionate about what they are doing.
Top comments (1)
Great read, Pradumna