No-slip

Inspiration

Our team was inspired by a simple frustration: swaps on Solana often return less than expected. Unexpected risks such as sudden volatility and shallow liquidity can cause unexpected slippage. We wanted to build a system that could see these potential loss before they happen and help users avoid bad trades. This led to creation of No-Slip: a tool that predicts slippage and optimizes swap timing in real time.

What it does

"No-Slip" continuously collects price data from Jupiter, converts it into structured bars, and feeds it into a multi AI forecasting engine. The system predicts whether the price is likely to rise or fall, identifies the best moment to execute a swap, and selects the safest route among Jupiter’s options. When the predicted optimal time arrives, it automatically executes the transaction on Solana using a signed transaction pipeline.

How we built it

We built backend with Python and TypeScript/JavaScript, and designed UI with Next.js. JupiterAPI and BirdeyeAPI was used to collect the data for prediction by http response. Once it gets the prediction, multiple LLM agents proceeds decision making through debate. Built multi agents for optimized strategy for finding optimal time period between the first predicted time and the first time that below than current price. Instead of linear regression to generate prediction, we used a Fourier transform based library named "Prophet" for generating prediction for future fluctuating target coin's price.

Challenges we ran into

Several challanges we ran into were deploying server, learning how to apply and stabilize APIs and dealing with inconsistent and noisy on-chain data. In addition, dealing with difference between development environments was also a challange.

Accomplishments that we're proud of

We are proud of our overall result done in 24 hours. We applied Jupiter API and Birdeye API for retrieving the data. We were also able to implement generative AI agent to improve decision-making.