Apr 24, 2025
With a passion for ZK and AI, Taipei-based student developer Albert Cheng sought to bridge Web2 social platforms, like X, with Web3 automation, leveraging a privacy-first workflow that verifies identities, links social activity, and distributes rewards automatically.
Enter TaskVault_AI.
TaskVault, built on Self for identity verification, enables projects to run campaigns where only real, verified users can complete tasks and earn rewards. No personal data is exposed, and everything is automated onchain. For example, a campaign might ask users to submit a proposal idea, answer a product-related prompt, or react to a governance update, with points awarded only to real humans who meet criteria like age, country, or uniqueness - no bots, no farming, no KOLs.
Below, we caught up with Albert for a quick Q&A to recap TaskVault & “Why Self?”
What inspired you to develop TaskVault?
We saw that protocols lacked fair, transparent ways to run campaigns without relying on key opinion leaders (KOLs), and real users didn’t have privacy-preserving methods to earn rewards. TaskVault solves both problems.
When did you decide identity verification was a core pillar?
From the very beginning—I wanted to solve privacy issues and prevent fake accounts right from the start.
Why did you choose Self for the identity layer?
Self is easy to integrate, straightforward for users, and lets us add privacy features quickly.
How is TaskVault made?
TaskVault AI combines zero-knowledge proofs, AI/ML models, and blockchain automation into a cohesive stack. Below are the core components:
Identity Verification Layer zkEmail Integration: Used zkEmail circuits to verify X account ownership via password reset emails. Generate proof by email file and send it onchain to verify and connect the address.
Self Protocol: Deployed a decentralized identity registry on Ethereum (SubmissionRegistry) to store verification requirements.
AI Evaluation Layer (ELIZA OS) NLP Pipeline: Built with ElizaOS Framework and integrated with X scraper to listen to and respond, and send rewards instantly on X. Projects can set their stylehash onchain and style to AI agents to prevent privacy and give judges different aspects to think and discuss.
Blockchain & Smart Contracts Core Contracts:
TaskVaultCore = "0x8441c3b1e6747605ab04e8a64f309bfba1fd37fe"
SubmissionRegistry = "0x0ff023acedbf133b998c70dd73d7b54db926cd44"
PrizeVault = "0xaf9048dca78acb1e4ea4a51af7b9e496a10c9e25"
ERC20Mock = "0xa5839608ff9511b66ab177530bd508727ca3455a"
Reward Distribution: Based on AI judge directly send point to user
Frontend & Integrations dApp Interface: Built with Next.js + TypeScript, using Privy for social login and wallet integration.
What's your vision for the future of TaskVault?
My vision is for TaskVault to set the standard for privacy-first engagement, letting everyone join campaigns and earn rewards with confidence, knowing their personal data is always protected and securely verified.
Learn more about Self
Start with our Quickstart docs here.
Let's talk
Self is a privacy-first identity verification platform designed for Web3 and universal apps.
Securely onboard Web3 users with private, trustless identity verification.
Enable Sybil-resistant polling for fair and transparent governance.
Check sanction lists to ensure compliance without exposing user data.
Facilitate wallet recovery with secure identity proofs.
