Our First AI x Web3 Hackathon Was Epic!
Hi there, it’s the EPIC AI NEWSLETTER. We didn’t post anything on Monday, and we have a good reason why!
Last weekend we hosted our first AI x Web3 Hackathon – dAGI House. Powered by cyber•Fund and Epic Web3.
We explored how to use AI to make crypto simple and widely adopted, and how to leverage Web3 building blocks to ensure AI remains unbiased and decentralized.
Here are some highlights:
Over three days, our venue welcomed more than 150 participants.
Mentors from top companies like Gagra VC, Galadriel, Corcel, Naptha AI, MBD.xyz, 0G labs, NEAR, Nevermined, Marlin, Chainbase and Phala Network conducted workshops and supported our participants throughout the hackathon.
Best of all, they stay in touch after the event. 💙Participants joined us from all over the world, including Europe, US, and Asia.
Our hackers were coding nonstop at the venue to make the submission deadline.
By the end of the hackathon, we received 40+ project submissions on DoraHacks!
There were many outstanding participants and here’s what our winners created!
1st Place: WarpAds
WarpAds is an open display ads protocol for Farcaster, analogous to the prominence of Facebook ads. This innovative platform allows users to convert any existing Farcaster frames into a WarpAds Frame (an ad-embedded frame) and relay the frame requests to the original server without requiring any changes on the server.
How It Works:
For Influencers: Influencers on Farcaster can create a WarpAds URL for popular frames they curate and start casting on WarpCast. As these frames are used by many people, higher click-through rates generate more revenue for influencers. The rendered WarpAds URL displays the original frame along with an embedded advertisement.
Ad Creation and Personalization: Anyone can create ads on the platform, which are labeled using Phala's AI coprocessor. These labels are utilized by MBD's content moderation APIs to offer personalized ads to users viewing a WarpAds frame.
Revenue and Royalties: Interaction data is tracked to compute ad click-through rates, rewarding influencers from the protocol's revenue. Original frame authors can redeem their royalties by setting an additional WarpAds metadata on the frame server to verify and claim their share of the ad revenue.
Technologies Used:
Phala Network: Ad labelling powered by AI agents. Decentralised Oracle compute in TEE to resolve influencer and author rewards
MBD: Content moderation for personalized ads.
Development Stack: Built on NodeJS, Express, Next.js, Solidity, and TypeScript.
and more.
🔗 Explore the details, and watch the demo: https://dorahacks.io/buidl/14055#details
2nd Place: AuthMark
“Detecting Synthetic Content is hard, deepfakes and voice cloning are becoming indistinguishable from real, these are powering a new wave of sophisticated scams and crimes.
Misinformation and fake news are becoming a global problem, and the current solutions are not enough to stop it.”
Features:
AI digital watermarking (methods for video, audio, images and documents)
Secure font "AuthMarkFont" (digital watermarked fonts)
Decentralized Proof of Life and Authenticity Protocol (dePoAP)
Reverse proxy for dynamic content delivery and status updates
Proof of Life & Proof of Authenticity should be owned and controlled by users, not big tech. Firewalls and Anti-virus software are important tools to protect against cybercrime.
Authmark is positioning itself to be a must-have tool to protect businesses and citizens against the next generation of malicious AI threats.
🔗 Explore the details, and watch the demo: https://dorahacks.io/buidl/14012
In an epic twist, dAGI House had four projects tied for third place
3rd Place – Jarvis
Jarvis is a Collaborative Token-Gated Pragmatic Multi-Agent.
Collaborative: meant for teams (eg traders)
Token-Gated: only those who own a Jarvis NFT with a specific number can access it
Pragmatic: It can build transactions to be sent on chain
Multi-Agent: It combines many agents (potentially an infinite number of agents) to get the most powerful AI assistant ever
🔗 https://dorahacks.io/buidl/14019
3rd Place – Task Llama
Imagine a world where artificial intelligence (AI) agents and models can be put to work on automating boring jobs (Craigslist for AI Jobs). Task Llama is a marketplace connecting task suppliers with AI agents, to automate any boring, repetitive, data-intensive tasks.
As a task supplier, you post tasks, provide a reward payout and have them completed through a combination of human and AI collaboration. To ensure the quality of the work, you can validate task completion using a variety of methods:
LLM critic benchmark
Static test cases (evaluation metrics)
Point & click demo
Crowdsourced community feedback (similar to platforms like lmsys)
🔗 https://dorahacks.io/buidl/14015
3rd Place – Open Market
Open Market offers a dynamic platform where users can buy and sell datasets, machine learning models, and human labelling services. Key features include:
Ditch traditional logins and embrace cryptographic keys and blockchain wallets for seamless, decentralized access.
Web3 Payments: All transactions are conducted using Ethereum, ensuring secure and efficient payments.
User Uploads: Users can freely upload their datasets and models. Upon sale, they receive payment minus a commission fee to the marketplace.
Human labelling: Sometimes data labelling needs human power. On the marketplace, users can pay or be paid by others to label data.
Chat with your data: Before Users buy a dataset they have the option to chat with a chatbot. This chatbot can answer all the information about the dataset without leaking any information. This gives users the guarantee to only buy what they want.
🔗 https://dorahacks.io/buidl/14031
3rd Place – Facterless
Facterless is a platform for scoring the accuracy of opinion leaders' predictions. It utilizes the power of open-source Galadriel blockchain technology ensuring data is immutable, with no X posts deleted. Custom AI-driven linguistic models process a large amount of data of X posts, cross-referencing these predictions with actual market outcomes provided by CoinMarketCap.
These features evaluate whether crypto influencers were correct in their predictions about market trends and calculate the average accuracy score from 0% to 100%, providing users with clear explanations of information that is behind each prediction assessment.
🔗 https://dorahacks.io/buidl/14027
📌 Explore the projects created by our hackers over these three days!
Thank you to all participants for their inspiring projects, and we encourage you to continue developing and innovating!
A special shout-out to our mentors for their invaluable support and guidance throughout the hackathon, sharing their expertise and helping our hackers excel.
And of course, a big thanks to the Epic team for making this hackathon possible!
Stay tuned for more exciting announcements coming soon. Follow our hackathon’s Telegram channel and Twitter to be the first to know!
That wraps it up for today! 👋 But before you go...
Check out our LinkedIn, Facebook or Twitter pages for more details. Follow us to stay updated on all the latest news!
Best,
Epic AI Dev team.