Livepeer-Community-Node / Grant-Program

Livepeer Community Transcoder Node's Grant Program
10 stars 5 forks source link

Tali AI - Discord technical / community support bot #26

Open niznet89 opened 1 year ago

niznet89 commented 1 year ago

Give a 3 sentence description about this proposal.

We aim to design a private Discord bot or Docs widget for Livepeer, leveraging cutting-edge Large Language Models (LLMs) to interlink Livepeer's Github repositories, documentation, articles, and Discord Q&A history. This interactive platform will provide a medium for users to pose queries in natural language, enhancing the Developer Relations, Customer Support, and Community Support experiences. The solution is devised to support Livepeer's teams, streamline Discord support hours, and expedite responses to frequently asked questions.

Describe the problem you are solving.

The major challenge in technical Web3 projects is that despite comprehensive documentation, developers and users often encounter questions that are not directly addressed in the available resources. Developer Relations and Community Support teams provide crucial support in these cases but face constraints due to the high volume of daily requests and questions. The issue here is the lack of an efficient, autonomous solution to handle routine inquiries, allowing the human teams to focus on unique customer issues and community engagement.

Describe the solution you are proposing.

We propose the development of an AI-Powered Discord bot or a Docs widget, integrated with Livepeer's various information sources like Github repositories, documentation, articles, and Discord Q&A history. By enabling users to pose queries in a natural language, this bot/widget aims to answer frequently asked questions effectively. This will augment Livepeer's DevRel and Community Support teams, leading to more efficient problem resolution, building trust in the protocol, and nurturing a vibrant community.

Describe the scope of the project including a rough timeline and milestones

The project has two primary milestones, each spanning approximately three weeks:

Milestone 1: Implementation of Data Sources (3 weeks) – Post the discovery phase, we aim to integrate 3-4 high-value data sources into the bot/widget, refining file loading, index creation, and data source integration.

Milestone 2: Testing and Optimization + Deployment (3 weeks) – This phase focuses on refining prompt engineering, eliminating hallucinations, and making the bot/widget production-ready for deployment on Livepeer's Discord channel or Docs page.

Please estimate hours spent on project based on the above

Considering the complexity of the project, it's estimated that each milestone will require approximately 120-150 hours of work, resulting in a total of 240-300 hours for the full project. This includes time for development, testing, optimization, and deployment stages.

Grant funding

We are asking for $5000 USD in funding which will cover development costs and service of Tali for a year.

Other notes

We have presented to the Livepeer community (on a community call) to receive feedback on the value of Tali within the community.