In a world packed with information, our dedicated team, consisting of Mónica Barroso, Marc Sánchez, David Ibañez, Glencora del Pino, David Creus, and myself, embarked on an exciting journey during Innovation Days. Our mission was to investigate a way to improve Fluendo's documentation by generating summaries of individual pages or sections using Large Language Models (LLM).
Imagine a scenario where Fluendo could use the power of this type of artificial intelligence to help with its internal documentation. As an example, LLM could be employed to create a summary of a page called "Getting Started with LLM" including the following information:
- Understanding LLM and its capabilities.
- Practical applications of LLM within the company.
- Advantages of employing LLM.
- How do you get started with using LLM at Fluendo?
New coworkers could read this summary to get a quick understanding of what it is and how it can be used; this would save them time from reading the entire page, which can be quite long and detailed.
But we don't stop here; we might also use this model to generate summaries of individual sections of its pages. For instance, it could create an overview of the "Fluendo Company" section within the "Welcome to Fluendo" page, focusing on essential aspects like:
- First hours in Fluendo.
- Fluendo’s history overview.
- Or Fluendo’s products and services summary.
For new flunions, such a summary would serve as a handy refresher, saving them time from having to re-read the entire section.
The benefits of LLM-based documentation summaries could be a great way to improve. They promise not only to enhance accessibility but also to keep users in the loop about the latest document changes. A weekly summary could provide insights into modifications made to specific pages like:
- What new sections were added to the page?
- What existing sections were updated?
- What are the most significant changes?
As we envision the future, a few potential applications of this LLM-powered summarization feature come to light:
- Automated generation of summaries for all new and updated pages, ensuring that users always have swift access to the latest changes in documentation.
- Empowering users to request targeted summaries of specific pages or sections is very helpful for those seeking quick insights on specific topics.
- Integration of the LLM-powered summarization feature into the search engine simplifies the process of finding relevant summaries when searching for information.
- A multilingual dimension, enabling documentation summaries in different languages, making Fluendo's information accessible to everyone.
In conclusion, the LLM-powered summarization feature could be invaluable to Fluendo's toolkit. It promises resource savings, ensures top-notch documentation quality, and amplifies the overall usability of Fluendo's documentation, a win-win for all. If you are interested in getting to know more about this project, contact us today.