The problem Flow Bot solves

The rapid growth of Web3 as a multi-chain ecosystem has introduced numerous new concepts and frameworks with each new growing chain, leading to a challenging and time-consuming learning curve for individuals aiming to adopt and build within these ecosystems. However, the existing support infrastructure within the Web3 ecosystem isn't built to support personalized doubts due to lack of community bandwidths There are community channels available but reaching out to those for smaller issues looks inappropriate. Moreover, the abundance of learning resources available often overwhelms learners, leading to burnout and information fatigue, hindering their ability to effectively navigate and utilize the available resources.

Therefore, there is a critical need to address these challenges by designing an efficient, user-centric knowledge acquisition and support system. The system should facilitate an easy learning process by providing ready access to relevant/useful information and doubt support. And it should have fast accessibility so that users can save their searching time and spend more time on buidling.

We developed a bot called Flow bot which is a LLM and has been trained on all of Flow documentation. And we shipped it as a vscode extension. Developers who are developing on flow blockchain can add this extension to their vscode and once enabled they can use it as a developer partner to get information about the concepts they are implementing, any doubts they have while developing, and many more. They don't have to locate anywhere just with a few shortcuts they can get access to the information they need.

Developer can ask questions like:

• What is consensus committe Distributed Key Generation Protocol ?
• How to the architecture of Flow accounts different from other chains accounts ?
• Can you share the URL where flow mainnet is accessible ?
• Can you explain me how can i write a hello world contract in cadence explain with some comments as well ?

Challenges we ran into

Challenges we ran:

• It was pretty time-consuming to preprocess the whole of flow documentation into LLM-understandable documents.

• We faced a few developmental issues while integrating LLM in vs code but with time we went through it.