Project Overview
CAREL Protocol is a privacy-first DeFi execution layer on Starknet — combining private execution (Hide Mode), autonomous AI trading agents, and a full-featured DEX.
Key Performance Indicators
These are the growth targets CAREL is tracking through the testnet and seed phases.
User Metrics
| KPI | Seed (3 months) | Growth (6 months) |
|---|---|---|
| Active wallets | 500 | 2,000 |
| Monthly active users | 200 | 800 |
| Registered referrals | 150 | 600 |
Transaction Metrics
| KPI | Seed (3 months) | Growth (6 months) |
|---|---|---|
| Total on-chain transactions | 5,000 | 25,000 |
| Hide Mode transactions | 1,000 | 8,000 |
| AI agent executions | 500 | 3,000 |
| ZK proofs generated | 1,000 | 8,000 |
Protocol Metrics
| KPI | Seed (3 months) | Growth (6 months) |
|---|---|---|
| Cumulative swap volume | $50K | $300K |
| Total value staked | $20K | $100K |
| Points epochs completed | 3 | 9 |
| Airdrop claims processed | 100 | 500 |
User Acquisition Strategy
CAREL’s user acquisition focuses on three channels:
1. Starknet Community
- Active presence in Starknet Discord, Telegram, and developer forums
- Co-marketing with Starknet ecosystem projects
- Participating in Starknet Foundation campaigns and ecosystem events
- Targeting existing Starknet users who already have wallets and testnet tokens
2. Crypto-Native Privacy Users
- Positioning Hide Mode as the primary differentiator for privacy-conscious DeFi users
- Content marketing around wallet unlinkability — explaining the real-world privacy benefit vs. normal on-chain transactions
- Targeting users active in privacy-focused communities (ZK, privacy tech)
3. Testnet Incentive Programs
- Points system with leaderboard drives repeat engagement
- Referral system gives users an on-chain incentive to bring others
- Testnet airdrop campaigns convert passive participants into active traders
- Discount Soulbound NFT creates a loyalty mechanism that rewards early adopters
4. Developer and Builder Channel
- Open-source codebase — developers can review, fork, and build on CAREL contracts
- ERC-8004 AI executor standard is a novel primitive — potential for third-party integrations
- Hackathon presence (Starknet ecosystem) generates developer-led word of mouth
Business Model
CAREL Protocol generates revenue through on-chain fees collected by the FeeCollector contract and routed to Treasury. The v6 economic model separates fee types clearly — swap fees go to dev/ops in USDC, AI fees use a hybrid 3-component model, airdrop fees are partially burned.
Protocol Fees
| Fee Type | Rate | Routing |
|---|---|---|
| Swap fee | 0.2% | 100% USDC → dev/ops |
| Bridge fee | 0.4% | 100% USDC → dev/ops |
| AI execution — L2 | 2 CAREL flat burn + $0.30 in CAREL (TWAP) | See breakdown below |
| AI execution — L3 | 3 CAREL flat burn + $0.50 in CAREL (TWAP) | See breakdown below |
| Airdrop claim fee | 5% | 2.5% burn + 1.25% treasury + 1.25% dev |
| NFT upgrade (CAREL) | Variable | 50% buyback-burned + 50% treasury reserve |
LP fees are separate and flow to liquidity providers, not the protocol treasury.
AI Fee Distribution (Hybrid 3-Component Model)
The v6 AI fee is a hybrid of three simultaneous components per transaction:
① Flat burn — 2 CAREL (L2) or 3 CAREL (L3) → 100% burned immediately. This guarantees a deflation floor regardless of CAREL price.
② USD-target revenue — $0.30 (L2) or $0.50 (L3) worth of CAREL, calculated via TWAP oracle at execution time:
- 75% → treasury accounting (stable USD value captured)
- 25% → weekly buyback fund
③ Buyback-burn epoch — the 25% buyback fund accumulates weekly, then treasury buys CAREL from the open market and burns it. Burn pressure scales with adoption volume.
Other Fee Distribution
Airdrop claim fee (5% total):
- 2.5% → Burned
- 1.25% → Treasury
- 1.25% → Development fund
External Revenue Anchors
Beyond on-chain protocol fees, CAREL targets three external revenue streams that reduce circular dependency on internal volume:
| Source | Model | Target |
|---|---|---|
| B2B API | Other protocols pay CAREL to access the AI execution layer | Active month 6+ |
| Treasury yield | 35% USDC reserve deployed to zkLend/Nostra (~4–6% APR passively) | Ongoing at TGE |
| Cross-chain fee sharing | Revenue from external bridge aggregator traffic | Growth phase |
Estimated combined contribution: 200–500K CAREL/month equivalent without depending on user transaction volume.
Revenue Sustainability
The v6 treasury holds multi-asset reserves — 35% USDC for operational stability, 35% CAREL for protocol operations, 20% ETH/BTC as a low-correlation hedge, and 10% LP positions for fee income. This gives the protocol 18+ months of operational runway even if CAREL price drops 80%.
Fee parameters are adjustable via on-chain governance (Governance + Timelock) after mainnet. The 48-hour timelock minimum enforces community visibility before any parameter change takes effect.
At testnet stage, CAREL does not yet generate real revenue. The business model is validated on-chain but operates on testnet tokens. Mainnet deployment is the prerequisite for real fee collection.
Project Cost Components
Breakdown of how the Starknet Foundation seed grant ($25,000) is allocated:
| Category | Allocation | Details |
|---|---|---|
| Development Device | 8% ($2,000) | MacBook M5 24GB — local Garaga Honk proof generation and full-stack testing |
| Infrastructure | 22% ($5,500) | Railway backend hosting, RPC node access, IPFS/Filecoin storage |
| Technical Development | 30% ($7,500) | V4 verifier redeploy, gas optimization, circuit binding, Sumo Login integration |
| Feature Completion | 20% ($5,000) | Shadow Bridge testnet, Hide Mode UX polish, AI agent improvements |
| Community Building | 20% ($5,000) | Discord/Telegram setup, ambassador program, referral campaigns, content |
Infrastructure Detail
| Item | Purpose |
|---|---|
| MacBook M5 24GB | Local Garaga Honk proving (24GB unified memory sufficient for ~16GB proofs) + development |
| Railway Pro (backend) | Rust/Axum relayer + privacy API, persistent uptime |
| RPC access | Starknet Sepolia node access for indexing and transaction submission |
| IPFS / Filecoin | NFT metadata storage (Discount Soulbound tiers) |
The MacBook M5 with 24GB unified memory replaces the need for a dedicated cloud proving server. Apple Silicon’s unified memory architecture is efficient enough to run Garaga Honk (UltraHonk) proofs locally, reducing recurring infrastructure costs significantly.
Seed Grant Milestones (3 Months)
The Starknet Foundation Seed Grant covers a 3-month execution window. Three sequential phases: foundation stabilization and security hardening, feature expansion and user acquisition, then scale and mainnet readiness.
Milestone 1 — Foundation & Stabilization ($10,000 — July 2026)
- MacBook Pro M5 24GB procured — ZK proving workload runs fully locally (required for Garaga Honk proof generation ≥16GB RAM)
- ShieldedPoolV4 verifier redeployed with
public_inputsbinding fix — live and verifiable on Starknet Sepolia - All 4 Noir circuits updated with full on-chain field constraints and private-exit token/amount binding (all AUDIT.md open items closed)
- AIExecutor gas reduced from ~5M to ≤1.5M — tested and verified on Sepolia
- TwapOracle gas reduced from ~3.5M to ≤200K — tested and verified on Sepolia
- Garaga Hide Mode gas optimized — minimum 30% reduction on Honk proof calldata path
- BTC Light Client PoW validation fully implemented in
btc_light_client.cairo(TODO stub replaced) - Mixing-window constraint moved from UX-only to on-chain enforcement in ShieldedPoolV4
- Hide Mode UX updated with real-time proof progress indicator, error states, and loading feedback
Milestone 2 — Growth & Integration ($8,500 — September 2026)
- Battleship Noir circuit (board commitment + move proof) and Groth16/BLS12-381 verifier deployed on Sepolia and callable from frontend
- Sumo Login smart contracts (SumoLoginContract + SumoAccountContract) deployed on Sepolia — OAuth JWT → ZK proof → Starknet smart account flow works end-to-end
- Sumo Login frontend tab (Google/Discord/Apple) integrated into wallet-connect dialog with OIDC backend handler
- AI agent interface live: natural language input, toggle L1/L2/L3, semi-autonomous mode, real-time execution plan display
- Leaderboard and referral UI live — real-time ranking, referral link generator, epoch countdown
- Battleship game UI fully functional on frontend — interactive board, move input, ZK proof status
- UI mobile-responsive at 375px viewport for trading, referral, and Sumo Login flows
- Galxe campaign launched — minimum 1,500 wallet completions achieved
Milestone 3 — Scale & Mainnet-Ready ($6,500 — November 2026)
- AI agent fully autonomous (ERC-8004 compliant): multi-step execution without per-step approval, live on Sepolia
- Shadow Bitcoin Bridge: BTC/WBTC lock-mint contract wiring complete with deposit flow, proof status, and claim interface functional on frontend
- Admin single-key migrated to Multisig + 48h Timelock on Sepolia
- Backend API hardened for 10K user load — rate limiting, auth middleware, Rust/Axum error handling validated
- Private executor stable under 10K active wallet simulation (retry logic, queue management, dead-letter handling)
- Zero critical open bugs — all GitHub issues from testnet community resolved
- Zealy sprint complete — minimum 800 engaged users
- Cumulative testnet campaign: 10,000 wallet interactions achieved
- Protocol in production-candidate state for Starknet mainnet deployment
After Seed — Growth Phase
Mainnet deployment, token launch, and the following features are Growth phase targets — not part of the seed grant scope.
Growth Phase Features
| Feature | Description |
|---|---|
| Social Media Tasks | On-chain rewards for privacy-related social tasks and content |
| CEX Deposit | Direct deposit from centralized exchanges into CAREL protocol |
| QRIS / Local Bank | Indonesian QRIS and local bank transfer onramp into Starknet |
| External Audit | Full third-party audit of 35 Cairo contracts, FE, and BE |
| Mainnet Deployment | Full contract redeploy on Starknet mainnet |
| CAREL Token Launch | Token live, fee collection active |
| Initial Liquidity | Protocol-owned liquidity seeded from 10% listing allocation |
| Governance Live | Timelock-enforced parameter changes operational |
QRIS and local bank integration targets the Indonesian and Southeast Asian market — enabling direct fiat-to-DeFi onboarding without CEX intermediaries.