TL;DR

This weekend marked a validation milestone for our team as
CopilotKit soared past 10,000 stars on GitHub, a testament to the path we are paving within AI developer tools. It has been quite a ride, and we are seeing the fruits of our development labor.

Starting a company is no easy task and the challenges and hurdles often seem endless. But anything is achievable with the right vision and mindset.

So how did we get here?

Last year, CopilotKit's CEO Atai, a former Meta engineer, started building an AI-copilot as a personal tool.

At the time, AI assistants, autocomplete features, and in-app chatbots were becoming increasingly common in developer tools like GitHub Copilot. Atai saw that this design pattern - a ChatGPT inside your application would be valuable everywhere.

Joining forces with Uli, our project quickly caught the attention of developers on GitHub and steadily started building an vibrant community of early adopters.

For fun, we decided to open-source the infrastructure we created to power the in-app Copilot and it immediately resonated.

We named it CopilotKit, and developers loved the elegant & easy to use infrastructure it provided.

The buzz was just the beginning!

Gaining Momentum

We presented CopilotKit at the AI Tinkerers event in Seattle, where it received a lot of attention. We were simply blown away and the positive response encouraged us to fully commit to developing this project further.

Currently, CopilotKit is designed to work with React applications, and we're planning to expand to other frameworks in the future. It also supports features like recursive agents through LangChain.

A Community-Led Journey

The Community feedback and collaboration have shaped CopilotKit every step of the way and it's because of them that we lept from 80 stars to over 10,000.

Our users have not just adopted the tool; they have helped shape it.

A Brand New Level - Collaboration with AI Intelligence

As we continue to innovate, we are continuing to break new ground in the Human-in-the-Loop (HITL) approach. During the build phase, human intelligence is applied to develop the agent’s cognitive framework, detailing the necessary steps and thought processes for completing specific tasks.

This integration of human insight into the AI architecture ensures that our tool, CoAgents, can monitor everything the agent is doing, while the agent can observe everything happening inside the app—all with just a single line of code. The agent’s state streams in real time as it updates.

Looking Forward: A Thank You to Our Users

We're excited to keep pushing the boundaries of what CopilotKit can achieve, aiming to make development smoother and more intuitive.

Please show your support ⭐

Come build the future of AI with us, thank you!

As a side note, when I started writing our story we had just crossed 10K stars on GitHub. Thanks to the incredible support we've already grown passed 11K stars at the time of publishing.