Loading...
Supporting developers, startups, and full-fledged companies building on Stellar and Soroban
© 2025 Stellar Development Foundation
Subscribe to the Newsletter
By Sava
The problem:
Turkish Lira lost over 80% of its value against the dollar in the last 5 years
Inflation is above 40% — bank deposit rates don't keep up
Savings lose value every month if held in ₺
Most Turkish people want USD-pegged stablecoins to protect themselves
Why they don't:
Doing it properly today requires four separate things:
A centralized exchange account (KYC, bank transfer)
A self-custody wallet (install app, back up seed phrase)
A DeFi protocol (research, trust, manual interaction)
The technical know-how to connect all three
Most people give up at step one and watch their savings erode.
Sava replaces that entire stack with one mobile app:
Sign in with email (no seed phrase, no separate wallet)
Deposit lira with a card (Visa, Mastercard)
Earn USDC yield on Stellar via Blend v2 lending pools
View everything in ₺ (Turkish Lira)
Sava composes three SCF Integration List building blocks end-to-end on Stellar — no off-chain ledger, no bridged asset, no custodial layer.
🔐 Privy — embedded Stellar wallet
Email, Google, or Apple sign-in → self-custodial Stellar account inside Privy's TEE
No seed phrase, no separate wallet app, no XLM balance required to start
SEP-10 binds the wallet to authenticated backend sessions
signRawHash flow signs Stellar transaction hashes client-side — wallet provider never sees transaction payloads
Result: kills the single biggest onboarding barrier for Turkish users
💰 Blend v2 — USDC yield on Soroban
Sava supplies USDC into Blend v2 pools (~$97M TVL across the protocol)
SupplyCollateral for deposits, WithdrawCollateral for withdrawals — yield accrues from borrower interest
Supply-only — Sava never opens borrow positions on behalf of users
Circuit breaker halts deposits if backstop drops, b_rate drifts, or oracle diverges from a secondary source
5-rule pool whitelist: audit, backstop, oracle, asset composition, TVL
Users withdraw at any time — no lockup, no penalty
💳 Mercuryo — Turkish Lira on-ramp
Buy USDC on Stellar with TRY via Visa or Mastercard
KYC and AML are Mercuryo's responsibility as a licensed EU provider, not Sava's
USDC settles directly to the user's Stellar account in seconds via incoming PaymentOp on Horizon
HMAC-signed widget URL constructed server-side — secret never touches the client
Already verified end-to-end on Mercuryo production: 600 TRY → 11.6 USDC, settled to Stellar mainnet account
The result: A Turkish user goes from "I want to save in dollars" to "I'm earning yield on USDC on Stellar" in about 3 minutes — without a single bridge, exchange account, or seed phrase backup.
$125.0K
Early User Interest:
200+ beta users onboarded to the testnet MVP, surfacing real UX feedback that shaped the current build.
1,000+ verified signups on the getsava.app waitlist (live counter visible on landing page)
Zero existing Stellar consumer products in Turkey — Sava is the first mover in a 25M-user market
Working Testnet App: Full end-to-end flow runs on Stellar testnet today — email sign-in via Privy → Stellar wallet creation → Mercuryo deposit → Blend v2 USDC supply → live yield in ₺ → withdraw. Demo video walks through the complete Turkish-language flow.
Mercuryo Production Verified: End-to-end TRY → USDC mainnet flow tested successfully — real Turkish lira in, real USDC settled to a Stellar mainnet account. Confirms the on-ramp leg works in production today.
Native Team: Two Turkey-based software engineers building Sava because we live the problem ourselves — converting our own savings from Lira to USDC every month. Clear mission to bring Stellar-native stablecoin savings to Turkey's 25M crypto users.
Goal: A Turkish user signs in with email, deposits testnet lira, and earns USDC yield — full round-trip working on TestFlight and Google Play Internal Testing.
Allocation: $25,000 (20%)
Deliverable 1 — Email / Google / Apple sign-in with auto-provisioned Stellar wallet
Brief description: User opens Sava, sees a 3-screen onboarding, signs in with email or social, and lands on a funded testnet Stellar account. No seed phrase. No XLM purchase needed first. Privy handles wallet provisioning under the hood.
How to measure completion: Public TestFlight URL in submission. Reviewer signs in with Google in under 60 seconds and reaches the home screen without team assistance.
Budget: $5,500
Deliverable 2 — Lira deposit flow with Mercuryo sandbox
Brief description: User taps "Add Lira", enters ₺500, completes a sandbox card payment in the Mercuryo widget inside the app, returns to Sava, and sees ₺500 of USDC arrive on their Stellar account with a clear deposit confirmation.
How to measure completion: Reviewer completes a ₺500 sandbox deposit. Receipt links to a Stellar testnet TX hash on Stellar Expert. Three test deposits at ₺100, ₺500, and ₺1,000 each have verifiable on-chain hashes.
Budget: $5,000
Deliverable 3 — USDC yield deposit and withdraw on Blend testnet
Brief description: User taps "Earn", confirms, and watches yield accrue in ₺ on the home screen. Withdraw works at any time, full or partial. Sava operates supply-only on Blend v2.
How to measure completion: Reviewer completes deposit → yield → withdraw round-trip. Each leg has a Stellar Expert TX hash. APY in app matches Blend testnet rate within ±0.1%. Partial 50% withdrawal verified on-chain.
Budget: $6,000
Deliverable 4 — Turkish-first UI across the entire app
Brief description: Every screen, error message, and currency in Turkish. ₺ formatting (1.250,50 ₺). Turkish date and number formats per device locale. Manual EN toggle in Settings for SCF reviewers.
How to measure completion: tr.json covers all UI strings — CI script fails on missing keys. Localization screenshot gallery in submission shows zero English fallbacks across the 12 main screens.
Budget: $3,500
Deliverable 5 — Public TestFlight + Google Play Internal Testing
Brief description: Sava installable by reviewers and beta cohort via TestFlight public link (iOS) and Google Play Internal Testing track (Android). Crash reporting and product analytics wired.
How to measure completion: Two public install URLs in submission. Reviewer installs and completes the full deposit-yield-withdraw flow. Read-only crash dashboard URL shared.
Budget: $3,000
Deliverable 6 — Backend API: SEP-10 auth and Mercuryo URL signing
Brief description: Hono backend deployed to a public production URL. Endpoints handle SEP-10 challenge/verify (binds Privy-issued Stellar account to authenticated session), HMAC-signed Mercuryo widget URLs (signing key never touches client), and webhook receiver with idempotent dedup against Postgres.
How to measure completion: Public OpenAPI spec at api.savafi.app/docs. CI green on GitHub repo. Reviewer runs included Postman collection — all 4 endpoints return expected responses.
Budget: $2,000
Goal: Sava becomes a real product on testnet — depositor-safety circuit breaker live, three growth features (yield calculator, savings goals, transaction history) shipped, closed beta with 20 waitlist users running real flows. Mainnet preparation done quietly in the background.
Allocation: $37,500 (30%)
Deliverable 1 — 7 layer defense, 3 source circuit breaker, and production monitoring
Brief description: Pool whitelist enforced at app startup. Supply-only invariant CI-tested on every commit. Circuit breaker halts deposits if backstop coverage drops below 8%, b_rate drift exceeds ±2% over 5 min, or oracle diverges from secondary source by ±0.5%. Trips fire in-app banners, push notifications, and on-call alerts. Withdrawals always allowed. Structured logs and read-only dashboard track everything.
How to measure completion: Reviewer triggers any of the 3 conditions in staging via env override, observes deposits halt + alerts fire within 30 seconds. CI tests for all 3 monitor types pass on the PR. Read-only dashboard URL with 7+ days of metrics in submission.
Budget: $11,000
Deliverable 2 — Yield calculator
Brief description: User enters a target ₺ amount and sees yield projection at current Blend testnet APY across 1/3/6/12 months, with comparison to Turkish bank deposit rates. Live updates as Blend's pool APY changes. Turkish-localized.
How to measure completion: Calculator live in TestFlight + Play. Output matches Blend testnet APY within ±0.1% for any input. Bank-rate comparison sourced from TCMB published data.
Budget: $5,000
Deliverable 3 — Savings goals
Brief description: User sets a goal (e.g., "₺10,000 vacation fund"), sees a progress bar, gets push notifications at 25/50/75/100% milestones. Progress bar tracks on-chain testnet balance.
How to measure completion: Goal creation works in app. Progress bar updates after on-chain testnet deposit confirms. Push notifications fire at milestone thresholds, verified on real device. Goal state persists across reinstall.
Budget: $5,000
Deliverable 4 — Transaction history + portfolio view in ₺
Brief description: Every Stellar testnet transaction (deposit, yield accrual, withdraw, on-ramp simulation) appears in a Turkish-localized list with the ₺ amount at time-of-transaction. Portfolio view shows cumulative yield over time with a 90-day chart and principal-vs-yield breakdown.
How to measure completion: History loads in <500ms. Time-of-transaction ₺ amounts reconcile with FX feed (CoinGecko + Binance fallback). Chart matches on-chain testnet state cross-checked against Stellar Expert.
Budget: $5,000
Deliverable 5 — Closed beta with 20 waitlist users on testnet
Brief description: 20 Turkish users from the existing 1,150-person waitlist run end-to-end flows on testnet across a 3-week structured cohort. Goal: validate which screens confuse users and where copy needs rewriting. In-app issue reporting + weekly check-in surveys. No real money at risk — all flows on testnet.
How to measure completion: 20 users onboarded (Privy console verifiable). ≥15 complete a full deposit-yield-withdraw cycle on testnet. Public GitHub project shows all reported issues triaged with priority labels. End-of-cohort summary published with top 5 friction points and Sava's response to each.
Budget: $6,000
Deliverable 6 — Mainnet readiness gate (pool selection + signed go/no-go)
Brief description: Specific Blend v2 mainnet pool chosen as Sava's launch pool, scored against the 5-rule whitelist. Selection rationale published, signed by both founders. App is configured for mainnet behind a feature flag — switch happens at Tranche 3 launch.
How to measure completion: docs/pool-whitelist.md published in public repo with 5 rules + scoring matrix for ≥3 candidate pools. Selected pool's contract address committed to apps/mobile/src/config/mainnet.ts (gated behind feature flag). Signed go/no-go document attesting v7.0 final-tranche bars are met.
Budget: $5,500
Goal: Sava ships to mainnet — Mercuryo production lira flow + Blend mainnet yield + App Store/Play Store listings all go live together. Real Turkish users move real lira, earn real yield. Polished onboarding, settings, in-app feedback, and customer support are part of the launch package, not afterthoughts.
Allocation: $62,500 (50%)
Deliverable 1 — Public mainnet launch on App Store + Google Play (Turkey)
Brief description: Sava publicly listed and downloadable in Turkey on both stores. Full Turkish App Store + Play Store assets (screenshots, descriptions, categories, privacy policy, terms of service). Mercuryo production widget and Blend v2 mainnet pool both wired and live at launch — no half-state. App passes Apple and Google review on first or second submission.
How to measure completion: Public App Store URL (Turkey region) and public Play Store URL (Turkey region) both resolve to the Sava listing. Reviewer downloads on a real device and completes a full mainnet round-trip (small real TRY top-up via Mercuryo → USDC supplied to Blend mainnet → yield accrues → withdraw). All TX hashes on Stellar Expert.
Budget: $16,000
Deliverable 2 — Real Turkish lira flowing through Sava on mainnet
Brief description: Real lira deposits via Mercuryo settle to user Stellar accounts as USDC on mainnet, median time-to-settlement under 15 minutes. Real users earn yield on Blend mainnet pool. Withdrawals work to user wallets. Mercuryo production and Blend mainnet have been integration-tested end-to-end before public launch (small team transactions verifying production keys, real lira flowing both directions).
How to measure completion: Pre-launch verification: 5 team-only mainnet transactions (real lira → USDC on Stellar mainnet → Blend mainnet pool → withdraw → off-ramp where supported), all TX hashes on Stellar Expert.
Budget: $15,000
Deliverable 3 — Polished onboarding, settings, and in-app feedback
Brief description: First-launch experience includes 4 welcome screens explaining the product (savings problem in Turkey → how Sava solves it → security + non-custodial guarantees → ready to start). Settings screen with: language toggle (TR/EN), notification preferences, currency display preferences, account info (Stellar address, copy-to-clipboard), privacy policy, terms, contact support. In-app feedback button on every screen lets users send issue reports directly to a Telegram support channel with auto-attached app version and current screen context.
How to measure completion: Reviewer launches Sava for the first time on a fresh device, sees 4 welcome screens, completes onboarding. Settings screen shows all listed options, language toggle works (verified screenshots in TR + EN). Feedback button on Home screen opens a form, submitted form arrives in Telegram support channel with metadata attached (verified by reviewer test).
Budget: $10,000
Deliverable 4 — Public TVL + activity dashboard
Brief description: Live dashboard at dashboard.getsava.app showing real-time mainnet TVL, total TRY processed via Mercuryo, daily active wallets, and 30-day yield earned. Built 100% on Horizon + Blend RPC reads — no off-chain ledger.
How to measure completion: Public dashboard URL. Reviewer cross-checks displayed TVL against Stellar Expert in real time, numbers reconcile within 1%. Methodology page explains each metric's data source. Source code in public repo.
Budget: $10,000
Deliverable 5 — Turkish customer support + help center
Brief description: In-app help center with 12 Turkish-language FAQ entries covering the most common user questions (how to deposit, how yield works, how to withdraw, what is non-custody, what happens if Sava goes down, how to recover access, etc.). Telegram support channel staffed during business hours by both founders for 30 days post-launch.
How to measure completion: Help center live in app with all 12 FAQs in Turkish (linked to from Settings). Public Telegram channel link in app footer. 30-day support log published at end of tranche showing ticket count, first-response times, and resolution times — average and percentiles.
Budget: $11,500
We are two software engineers based in Turkey. We experience the problem Sava solves firsthand, we convert our own savings from Lira to USDC to protect against inflation, just like millions of other Turkish people do. We know the pain of juggling exchanges, wallets, and protocols because we've done it ourselves for years. Sava started as the app we wanted to use but couldn't find.
Tolga Yaycı — Co-founder & Software Engineer
Computer Engineering degree. Previous SCF grantee — Sora (Stellar developer toolkit) was funded twice by SCF, both delivered successfully. Has built and shipped products across Stellar, Near, Arbitrum, Solana, and Internet Computer ecosystems. Responsible for Sava's Stellar integration, Blend Protocol connection, and mobile app development.
GitHub: https://github.com/tolgayayci
LinkedIn: https://www.linkedin.com/in/tolgayayci/
X: https://x.com/tolgayayci
Mert Köklü — Co-founder & Software Engineer
Computer Engineering degree. Previously led an engineering team at an NVIDIA partner company. Has received and delivered grants from Arbitrum, Cardano, and Sia ecosystems. Experienced in SDK architecture, API design, and backend systems. Responsible for Sava's backend infrastructure, API layer, and on-ramp integration.
GitHub: https://github.com/justmert
LinkedIn: https://www.linkedin.com/in/mertkoklu/
X: https://x.com/devmertt
Sava is bootstrapped — we built the working beta with 200+ users without external funding and now starting the release of end version with SCF support.
Tolga Yaycı
No other submissions.