Velocity

Demo

About

The problem it solves 0G's impressive potential in scope is hindered by a lack of developer tooling and education. Velocity introduces a modern CLI Quick-Start to get users onboarded fast and introduces developers to the key features of the platform. We authored a comprehensive skills repository to inform a developer's AI agent on everything it needs to know, which ensures that 0G keeps up with industry advancements in agentic coding. The CLI also allows users to get started with technically complex features such as iNFT agents, AI inference, and fine-tuning within seconds, giving users a unique foothold when starting their projects. Finally, the CLI is mirrored with a web-based GUI sandbox allowing developers to learn about and play with the breadth of features 0G has to offer. The team behind Velocity wholeheartedly believes that the potential of 0G's platform can be further recognized by the product we have submitted. Our repository has an open-source license from MIT allowing your platform to integrate Velocity into your dev-tooling suite. Use of AI tools and agents Our team used AIs like ChatGPT for research and coding in several instances, including parsing documentation, creating/debugging code, making architecture decisions, and creating the logo. Agentic AI like Cursor and Claude were also used for creating/debugging code.

How it's made

Challenges we ran into The largest challenge for our team was creative direction. The project took many turns before becoming Velocity, and did not follow a traditional ideation route. The original idea was to make a private Captcha system using zk-Proofs, which solves the implications of traditional Captcha's such as centralized data collection of cookies, user telemetry, and the utilization of human inputs to train AI models without consent. We elected to use 0G's compute to power our AI model. By trying to use 0G's testnet faucet and failing due to various reasons (e.g. not enough twitter followers, unexplainable verification fails), we were inspired to create an improved way for developers to gain testnet tokens that promotes privacy first. So the idea shifted to more of a developer first perspective. We created a system for developers to solve a captcha puzzle, locally compute a zk-proof of their inputs, and show the verification of that zk-proof to the testnet provider. This solution solves the issue of privacy and maintains integrity from the testnet provider's point of view, making the system a win-win. The data colleced can be optionally added to the AI model we used for fine tuning at the consent of the user. We also wanted to bring this solution to the command line, which happened to nicely converge with the track about dev tooling for 0G. As such, we ended up creating a modern CLI that contains all of the projects we did on 0G as modules that other developers can start quickly with.