The problem Donna solves

Problem 1: Initial Learning Curve and Jargon Overload

Navigating the intricacies of the Web3 ecosystem proves daunting for the average user due to its high initial learning curve and an abundance of technical jargon. Donna resolves this issue by implementing Generative AI and Large Language Models, such as OpenAI, to transform complex transactions into comprehensible English commands. This strategic simplification enables users to seamlessly integrate into the Web3 space, overcoming the barriers posed by its distinctiveness from traditional economic systems.

Problem 2: Fragmented Features Across Web3 Applications

The absence of a unified application supporting all essential features within the Web3 landscape, where certain platforms specialize in specific functionalities like token transfers or exchanges, poses a substantial challenge. Donna addresses this fragmentation by providing a user-friendly chatbot interface, facilitating a comfortable and engaging interaction with the ecosystem. By offering a comprehensive platform, Donna caters to users accustomed to both Web2 and Web3, simplifying intricate procedures like token swaps.

Problem 3: Decision-Making Amidst Abundance of Alternatives

The sheer volume of alternatives in the Web3 environment complicates decision-making for users. Donna leverages Generative AI to harness extensive training data, enabling real-time calculation of smarter and faster decisions. The AI's advisory capabilities assist users by highlighting potential errors and offering insights, empowering them to make informed choices amidst the vast array of options.

Problem 4: Lengthy and Complex Contract Deployment

The cumbersome process of writing and deploying contracts across different chains presents a significant hurdle. Donna streamlines this process by utilizing Generative AI to accurately generate simple contracts. Automated pipelines are then established to efficiently fetch code and deploy it to the required chains.

Challenges we ran into

During the development of Donna, our AI wallet assistant powered by Generative AI, we faced several challenges across different stages:

Ideation:

In the ideation phase, we initially considered using a browser extension for its quick access and modular appearance. However, as we explored the required functionalities, it became apparent that the extension would be bulky and cumbersome affecting user experience negatively. Consequently, we shifted our focus to a web application for a more streamlined and efficient platform.

Frontend:

Frontend development brought challenges in parsing and adapting answers generated by the Generative AI to fit various formats across different blockchain networks. The integration of several libraries and APIs, some of which were new to our team, introduced additional complexities that required careful consideration in their implementation. Securing the transaction signing process in the frontend posed its own set of challenges .

Middleware:

Challenges encountered during middleware development included navigating the complexities of incorporating the relatively new LangChain library. Ensuring effective task delegation among frontend, middleware, and backend proved critical for sustaining a cohesive development workflow. Additionally, formulating seamless workflows to integrate the wallet manager frontend with the various networks supported by the project demanded meticulous planning.

Code Generation Server:

Designing prompts to generate appropriate smart contracts from user instructions proved to be a nuanced challenge in the Code Generation Server. Implementing an automated workflow for contract generation and deployment while prioritizing security and ensuring users' ownership of the contract posed intricate challenges. Balancing these aspects required a thoughtful and meticulous approach to uphold the integrity and functionality of the code generation process.