Negotiations today often rely on trusted intermediaries, risking sensitive data leaks or biased outcomes. Parties need privacy and fairness in automated negotiations. So, our solution aims to increase the trust.
We started to build on SP1. However, soon down the line we realised how scarce its documentation is and on top of that the prover requires a 128 GB of memory and the public Prover network is invite-only. So, we had to quickly shift to Risc0 on the morning of 8th December to get the project in place.
Apart from this, building AI agents was a bit of a hassle as well. LLMs are meant to be unpredictable, while the blockchain demands every bit to be specified. So, making both of them work together was not easy.
Tracks Applied (3)
Walrus
Socket
Polygon
Discussion