APPLICATION INTRODUCTION
We propose to develop a private Discord bot or Docs widget for Fantom, harnessing the latest Large Language Models (LLMs). This bot or widget, by interconnecting Fantom’s Github repositories, documentation, related articles, and Discord Q&A history, aims to offer users a conversational interface to ask questions in a natural language.
Purpose of the system
The primary purpose of this system is to augment Fantom’s Developer Relations (DevRel) and Community Support teams, lessen the support hours on Discord, and provide quicker responses to frequently asked questions.
Scope of the system
The scope encompasses integration with Fantom’s Github repositories, documentation, related articles, and previous Discord Q&A, enabling comprehensive responses to user queries.
Objectives and success criteria
The objective is to enhance the efficiency of the DevRel and Community Support teams and improve the user experience within the Fantom ecosystem. The success criteria include a reduction in support hours and faster response time.
References
Based on the established problem of managing numerous requests and questions by the support teams, the need for an AI-assisted solution has emerged.
Overview
The proposed bot or widget will serve as an AI-assisted solution for handling frequently asked questions, enabling the support teams to focus on unique customer issues.
TEAM
Ali Agha: A technologist and entrepreneur with significant experience in decentralized solutions. He will be leading the technical development of the project.
- Roles and responsibilities: Co-founder, Developer
- Social media handles: Twitter
- Relevant experience: Web3 consulting services for startups and corporations like IBM and Thomson Reuters
- GitHub repo: OlypsisAli
Tenzin Rose: An entrepreneur and full-stack developer with a background in enterprise sales. He will be responsible for integrating the proposed bot with Fantom’s system.
- Roles and responsibilities: Co-founder, Developer
- Social media handles: Twitter
- Relevant experience: Worked with global startups and enterprises, specializing in project delivery. Has worked and delivered across numerous web development projects.
- GitHub repo: niznet89
TOTAL BUDGET AND FUNDING TIER
We request a $5,000 USD grant to develop and maintain the proposed bot or widget. The allocation includes infrastructure costs (OpenAI token usage, Digital Ocean hosting) for 12 months and development costs, which may involve contracting Data Science/ML experts if needed.
CURRENT FUNCTIONALITY
We have a working implementation of Tali (MVP) but a lot of development work is still yet to be done to improve the quality of answers, ingesting different data sources and providing analytics / observability so Fantom’s team has insight into what Developers are asking.
Currently deployed in Lens Protocol & Balancer Protocol as a Beta program.
TECHNICAL PROPOSAL
The proposed bot or widget will be a complex integration of LLMs, data indexing, optimization tools, and user interface design. It will involve data collection from Fantom’s Github repositories, official documentation, and past Q&A from Discord. Python will be used for the bot’s back-end development, with integration of LLMs for understanding and generating responses to user queries.
Nonfunctional requirements
Nonfunctional requirements include usability, reliability, performance, and supportability of the bot or widget, legal and licensing considerations, and its implementation interface. The bot will be hosted on Digital Ocean, ensuring high availability and performance.
SYSTEM MODEL
The system will involve scenarios like user queries, retrieval of relevant information from data sources, and generation of responses. The use cases will include querying the bot on Discord or the Docs page. The object model will consist of user interface elements and a complex integration of LLMs and data sources.
DEVELOPMENT ROADMAP
- Pre-Implementation: Collaborate with DevRel, Customer Support, and Community Support teams to understand their needs.
- Milestone 1: Implement 3-4 high-value data sources into the bot or widget, refining file loading, index creation, and data source integration. (Duration: 1 week)
- Milestone 2: Test and optimize the bot, ensuring its readiness for production. Final delivery is the bot into the community. (Duration: 3 weeks)
MAINTENANCE CONSIDERATIONS
Regular maintenance, updates, and improvements will be part of our ongoing commitment. This includes regular checks on the bot’s functionality, debugging, and updates based on user feedback and evolving requirements.