We conducted our second Pieces Open Source Meetup on December 20th, 2023 on our Discord Stage.
During the meetup, we celebrated the community's remarkable growth, our successful Product Hunt launch, and the wealth of community contributions that have significantly propelled our SDKs and projects forward. These highlights not only reflect the dynamism of our collaborative ecosystem but also set the stage for the exciting developments we're eager to share as we continue to refine and integrate Pieces into developers' workflows globally.
You can find the recorded session on our YouTube channel.
Here are some of the major highlights in case you missed the meetup!
Community and Open-Source SDK Updates
We highlighted several Pieces for Developers updates from the past two weeks. One significant update is the "Bring Your Own Model" feature, which allows for more flexibility and expansion beyond specific models like GPT-3 or GPT-4. We also released updates to the Pieces Desktop App and Pieces OS, including the ability to use your own OpenAI API key. This feature allows users to integrate their fine-tuned models into the Pieces ecosystem, a development that is expected to be particularly beneficial for larger companies and enterprises.
Mason also mentioned a successful Twitter space hosted last Sunday, focusing on developer productivity and AI, and a livestream featuring the co-founders of Pieces. Lastly, we also had a successful Product Hunt launch, where we ranked second for the day and fifth for the week.
We have been diligently working on numerous updates, particularly for the generated runtime. A new version of the npm package (1.2.0) is now available, which includes several enhancements such as a new code text segmentation endpoint. This endpoint allows users to pass in code and text, segmenting them into two separate parts. We encourage developers to update their packages if they're building on Pieces OS. The Example TS project has also been updated to use the most recent Pieces OS version. Additionally, we have introduced a new UI for the assets and copilot in the Example TS project, which now displays three snippets at a time in a grid view. We are planning to add more issues around UI updates and metadata enhancements.
Shoutouts for Community Contributions
We have seen a huge rise in community contributions to our SDKs and example projects!
We’re so thankful for some of the amazing contributions that were made to the projects in the past two weeks, including—but not limited to—Dera Johnson, Agrim, and Jeremiah Boothe, for working on a range of contributions including UI, CI/CD, and the Kotlin SDK.
Sophia emphasized the importance of community contributions and provided guidance on how to get involved with our open-source GitHub repositories, particularly for newcomers through the tag, "Good First Issues." She also introduced the "Pending Reassignment" label, which is applied to issues when there's no communication from the assignee within three days, which encourages regular updates from contributors.
We also covered the Kotlin SDK's availability on Maven Central, expanding the project's reach to developers in the Kotlin and Java ecosystems.
The discussion moved to the importance of the discussion boards and the description of the "desk buddy" project, which aims to assist developers with their workflow.
Hands-on Pieces Copilot Example Demo
Caleb took us on a deep dive into the technical intricacies of our Pieces Copilot example.
He began by explaining the role of the conversation stream, a crucial component that allows developers to access all conversations within the system.
Next, he moved on to the question stream, another vital component facilitating interaction with the models. He demonstrated how developers can build a request, send it off to the relevant endpoint, and then use the output to ground the model, ensuring that the results are highly relevant to the query and the developer's workflow.
Caleb also highlighted the importance of the model API and the model download progress stream. These tools allow developers to choose a different model to use throughout the copilot or even download a local model. He emphasized the convenience of showing the progress of a download, enhancing the user experience.
One of the key features he discussed was the output model, which provides the status of a message and the output of the LLM. This model is crucial for developers to understand when a question is completed or if a message has failed. He also touched on the conversation ID feature, which was recently added when we introduced persistent conversations. This feature allows developers to uniquely identify each conversation, making it possible to have the same conversation open in different apps at the same time.
We will be adding more documentation and videos will be released on how to translate some of these functionalities into TypeScript (and other languages) in future sessions. This will further aid developers in understanding and utilizing the full potential of the copilot example.
The session concluded with a reminder for the community to keep an eye on our upcoming copilot example and also an open invitation to the community to contribute to Good First Issues on GitHub.
We are excited to welcome all the new folks who have joined the community recently and can’t wait to see what we can accomplish as a group, as we all work together on these great ideas.
Stay tuned for more updates and join us in our journey to make Pieces an indispensable part of the development workflow worldwide.