The problem TrustHire solves

The Trust Deficit in the Gig Economy

In today's digital economy, a massive amount of freelance work happens informally. Clients and talent connecting directly through social media, Discord, or WhatsApp. While this bypasses the massive 20% fees of traditional platforms, it creates a severe trust deficit with zero protection for either side:

• Freelancers constantly worry about clients defaulting on payments, or ghosting them entirely after the final work is delivered.
• Clients worry about hiring fake profiles, encountering fabricated portfolios, or receiving subpar work because they have no formal recourse or arbitration.
• Pay-to-play barriers: Even when users try to rely on traditional platforms, freelancers are often forced to buy virtual "tokens" or "connects" just for the privilege of applying to a job, and face manual dispute systems that are easily manipulated.

How TrustHire Solves This

TrustHire completely removes the need for expensive, subjective middlemen. It replaces them with trustless Web3 infrastructure and unbiased AI, providing institutional-grade safety for independent and direct hires.

People can use TrustHire to securely hire talent or accept freelance work with zero anxiety about getting scammed. It makes gig work radically safer through four core pillars:

1. Guaranteed Payments (Sui Escrow): Clients securely lock project funds in an on-chain Sui Move contract before work begins. The platform never holds custody of the funds, and the freelancer knows the money is guaranteed the moment they meet the requirements.
2. Unbiased AI Arbitration (Gonka Router): Instead of relying on a biased human support agent during disputes, TrustHire routes milestone submissions (like a GitHub PR or Figma link) through a multi-model AI consensus engine. The AI neutrally evaluates the work against the initial project spec, outputs a "Truth Score," and automatically authorizes the release of the escrowed funds.
3. Immutable Reputation: Freelancers build a permanent, on-chain work history that proves their skills and track record, completely eliminating the "fake profile" problem.
4. Zero-Cost, Walletless Onboarding: Unlike legacy platforms that charge freelancers just to apply, TrustHire removes all friction. Using zkLogin and Enoki Sponsored Transactions, talent can sign in with a simple Google account and start working instantly. Gas fees are fully sponsored, meaning freelancers don't even need a crypto wallet or SUI tokens to get started.

By combining Web3 payment guarantees with AI verification, TrustHire makes freelance work fairer, safer, and completely trustless.

Challenges we ran into

1. The Enoki Gas Sponsorship vs. Coin Splitting Hurdle

The Challenge:
Our goal was to make the platform 100% gasless using Enoki Sponsored Transactions. However, when we attempted to sponsor the create_escrow transaction (where a client locks SUI funds into the smart contract), the transaction kept failing. We discovered a core limitation: Enoki gas pools cannot sponsor transactions that involve splitting the primary gas-paying coin itself, which is exactly what happens when you try to extract funds to lock into an escrow.

The Solution:
We had to deeply rethink our transaction architecture. We leveraged Programmable Transaction Blocks (PTBs) to elegantly solve this. Instead of a clumsy two-step process, we used a PTB to atomically execute tx.splitCoins and pass the newly created coin object directly into the tx.moveCall for the escrow in a single execution. We then explicitly restricted Enoki sponsorship to the freelancer side (onboarding and milestone submissions), ensuring the talent experience remained 100% frictionless and gasless, while clients handle standard gas for their funding transactions.

2. Preventing AI Hallucinations in Financial Decisions

The Challenge:
Since our smart contracts release real escrowed funds based on the AI's verdict, relying on a single AI model to verify complex milestone deliverables (like analyzing a GitHub PR) was too risky. A single model might hallucinate or provide a subjective, lenient approval.

The Solution:
We implemented a strict Multi-Model Consensus mechanism using the Gonka Router. Instead of one model, we cross-reference the milestone proof across multiple distinct models (e.g., DeepSeek and MiniMax). Only when the models reach a consensus on the requirements does the system output a definitive "Truth Score." If the models disagree, the system flags it for manual review, ensuring that no funds are released on a hallucinated verdict.

3. Cost-Efficient Hybrid Architecture (Off-chain DB + On-chain Execution)

The Challenge:
Storing extensive project requirements, freelancer portfolios, and AI reasoning traces directly on the Sui blockchain would have resulted in astronomically high gas fees. Furthermore, we needed a way to render complex UI data instantly on page load without waiting for wallet connections or slow RPC nodes to hydrate the state.

The Solution:
We implemented a hybrid architecture to solve both problems. We use Supabase as a lightning-fast off-chain database to store all heavy metadata (project descriptions, profile text, AI analysis).

When a critical financial action occurs (like creating an escrow or approving a milestone), the transaction is executed on the Sui blockchain. We then capture the resulting on-chain transaction hash (txDigest) and store it directly in our off-chain Supabase database as an irrefutable reference.

This approach drastically minimized gas fees while giving us Web2 load speeds. We render the heavy UI from Supabase instantly, and use the stored transaction hashes to cross-verify the financial state on-chain exactly when needed.