At the heart of any new product development lies one critical component — the user. This reality was at the forefront of our minds as we updated Label Studio to the newest 1.8 release. These new features could be lost in the shuffle without our users' understanding and adoption.

Enter the Generative AI template library — a fresh take on adapting foundation models. This game-changing feature offered users the tools to easily modify models to their unique needs, be it for chatbot model assessment, fine-tuning Language Model Learning (LLMs), or image generation.

👀
View the blog post behind this work, here.

But it's not just the variety that's impressive; because of Label Studio’s flexible and open-source nature, the breadth of models that could be used with this is endless. We catered to a wide spectrum of user requirements: from the sophisticated Llama 2 and the powerful GPT-4 to the precise Whisper, the versatile Alpaca, and the adaptive PaLM 2.

Having an arsenal of top-notch features is one thing; enabling users to navigate and utilize these tools is another challenge. This is where my role took center stage. As experienced through previous successes, the key to driving user adoption was to have an in-depth yet accessible guide.

So, I rolled up my sleeves and created comprehensive educational content and documentation that cut through the complexity of Generative AI Templates. I aimed to build a bridge between our users and the product, simplifying the advanced technology to a language that resonated with all, regardless of their expertise level.

The goal was to foster an environment of self-guided learning where any user could confidently explore and harness the potential of these new features. The user feedback we received spoke volumes about the effectiveness of this approach. Users were not only able to understand the new features but also felt empowered to adapt and utilize them in their unique contexts.

Oh, I 😍 that you made both a blog post and a video! Different learning modalities for different folks (or in my case, two modalities for the same person, because my brain is greedy and learns best if I have both 😁).
Laura Langdon's reply to this guide — all about providing different modalities of learning.

This work wasn't merely about immediate user adoption but also about setting the course for our future product direction. We were ushering in a new era of HumanSignal — empowering further AI model adaptation. Getting the education piece right was essential.

I believe that when users are given the keys to understanding and controlling the technology, they become more than just users but instead become partners in shaping the future of the product. Working in open source, a tight feedback loop with your users, as well as plenty of accessible content that allows the user to be successful, is imperative. This was a particular highlight at Label Studio, truly leaning into this open-source ecosystem to build a community-driven tool that grows and evolves with its users.

In the words of John Mulaney, "You're all gonna die. Filled with regret." Regret for not trying new things and for not taking innovation’s reins. But not on my watch! With this guide, we've ensured that every user can confidently navigate the new features of Label Studio and pave their way into the future of AI models.

Project Recap

Primary Objective: Enable user adoption of new templates through educational content.

Secondary Objective: Support the company’s larger endeavors as they progress into further work in the Generative AI space.

Skills used:

  • Cross Team collaboration with Product Marketing, Engineering, and Customer Success
  • Technical Documentation
  • Video Content Creation

Tagged in:

Label Studio Machine Learning Python Data Science Past Work DevRel Education Tech Education All Posts