DatAgent DAO

AI has been trending all over the world be it chatGPT, stable diffusion, etc. These models are trained on data being served to them through internet, etc. Currently its a bunch of people who decide what data to serve and various other decision, giving them total control over the protocol. This has led to the concerns stated by many people around the world over centralized control of something like AI having immense capabilities in them. Don’t worry DatAgent has come to your rescue.

DatAgent DAO is one-stop complete on-chain solution for decentralizing the this whole AI protocols and giving control to users to improve as well as have a say in decision making of them. It provide protocol governance with help of DAO on FVM which allows on-chain dataset transfers along with the general features of DAOs enabling voting on what data the AI is going to be trained and what not. Moreover, anyone can put their own dataset as proposal to make the model more efficient and earn the governance token in return, this would motivate many people to provide dataset which would eventually improve the model a lot as well. People could also ask the protocols for certain datasets and models as per their ask so as to use them for themselves as well. This way the decision making in the AI protocol is decentralized but let’s say we have an approved dataset, is there a way to train the model and get the results throughon-chain transactions?

The answer is yes, thanks to Bacalhau and their project Lillypad. The models could be deployed on bacalhau and be interacted on-chain through smart contracts. Once the proposal of dataset is accepted one can initiate the training of model on that on-chain and even get results on that fine-tuned model on-chain with help of our DAO. This results in complete decentralized on-chain solution for complete working of an AI company enabling users to have a say in the protocol and having complete transparency on decision being made.

We have tried to make a complete on-chain solution thus have used a lot of decentralized protocols on our Dapp.

The DAO contract consists of deal client as well as the Bacalhau’s lillypad contracts and is deployed on FVM enabling on-chain deal making as well as on-chain interaction with the deployed AI models on Bacalhau. Direct interaction with models is also enabled with the help of Lilypad’s smart contracts. All the results displayed i.e. image retrievals, etc are done with the help of Saturn’s CDN built on Filecoin. We have not even used a centralized database in our dapp, thanks to Tableland, we have implemented a decentralized database to store the proposals, this enables users to have access of their data rather than making it public to everyone or trusting someone else with that.

Beryx API helped us to track how far the proposal has reached. By storing the transaction in Tableland, one can see the status of his transactions and proposals easily.

We have also enabled notifications for users about the decisions getting made on DAO and tracking one’s own proposal with the help of Push Protocol, with the help of data we get through the Beryx API.

As you could see, we have integrated various protocols so as to provide complete on-chain solutions to users, maintaining decentralism throughout which makes our project true use-case of core principles of blockchain and web3.