Algorand xGov-funded wallet Confío now settles on BNB Smart Chain

A Proposal, and Where It Settles Now

In late February, a two-person team asked Algorand's community-run xGov¹ grant program for 250,000² ALGO to scale Confío, an open-source, non-custodial mobile wallet pitched on the Algorand forum³ as 'Web2-to-Algorand Consumer Onboarding Infrastructure (LATAM)'. The proposal described a deployed mainnet stack — a React Native app, a Django/GraphQL backend on AWS, PyTeal/TEAL smart contracts, a USDC-a-backed stablecoin called cUSD, 5,500⁴ registered users and 70+⁵ funded users with USDC-a deposited on-chain (all figures per the proposal) — and billed the repository⁶ as 'Latin America's PayPal'.

Today the same project describes itself differently. The repository README opens with 'The current product system settles on BNB Smart Chain,' and confio.lat⁷ markets a yield-bearing dollar savings token backed by Ondo Finance⁸'s USDY — on BNB — alongside a $CONFIO community token, also on BNB. Algorand, the README explains, remains in the repo 'for historical verification and migration support.'

The route between those two descriptions: the first proposal won a heavy majority of vote weight — 471,133⁹ up against 4,499 down — but fell short of quorum and was rejected; the team resubmitted, was funded, and the payout cleared on May 25; a BNB Smart Chain cUSD+ vault was deployed on July 10; and the project's own whitepaper, dated August 10, now states the product 'settles entirely on BNB Smart Chain.' How a would-be Algorand onboarding layer worked the program, got paid, and ended up on another chain is the fuller story.

A Wallet That Hides the Blockchain

Confío's thesis is that Latin American stablecoin demand is not a crypto phenomenon but a dollar-access problem. Argentina's 2001 'Corralito' — the overnight freeze of every bank account in the country — destroyed a generation's trust in custodial finance, the proposal argues, and the people who most need stable dollars are WhatsApp users, not MetaMask users. 'They will not write down 24¹⁰ words.' Self-custody, in this framing, is not ideology but risk management learned from lived experience.

The design follows. Keys are generated on the user's device, with recovery material encrypted in the user's own Google Drive or iCloud, so the operator never holds private keys. A backend maps phone numbers to Algorand addresses so users send by contact name. Every action is a sponsored atomic group transaction: Algorand lets one account pay the network fee for another, and grouping transactions into an atomic bundle means the whole flow — transfer, fee payment, any contract calls — executes or fails as one unit. The user pays nothing and sees 'Sent $10¹¹ to María ✓' after roughly 3.5¹² seconds of settlement.

The project is led by Julian Moon, a Korean full-stack developer who says he has lived across Latin America for five years and built a Spanish-language TikTok audience the proposal puts above 430,000¹³ followers, with Susy Ramirez running content and a 12,000+¹⁴ member Telegram¹⁵ community (both per the proposal); the whitepaper later cites an audience of roughly 480,000¹⁶ across social platforms. The team also lists Algorand Foundation Accelerator participation, and confio.lat's domain was registered in March 2025. One repo artifact from that early period is a chain-decision note showing the codebase stored Algorand addresses in a database field named aptos_address while the founders weighed Aptos against Algorand before launch — the same pragmatic chain-agnosticism visible in today's move.

What is verifiable on-chain: the cUSD contract (app 3198259271¹⁷) shows 64,838 cUSD minted lifetime, 25,114 currently circulating against a matching 25,114 USDC-a lock — the 1:1 backing the proposal advertised, visible in the contract's own global state — and the contract is not paused, with the most recent real cUSD transfer on August 14 and sponsor-account activity as late as August 19. cUSD is not listed on Tinyman or Pact: it is a closed-loop app balance rather than a tradeable market. The site's own counters read US$72,696 deposited on-chain and US$3,983 raised in the $CONFIO presale, and the whitepaper's July 23 operating snapshot (8,004¹⁸ phone-complete users, 177¹⁹ government-ID-verified users, 2,092²⁰ devices used in the prior 30 days) is labeled by the project itself as 'internal operating metrics rather than independently audited figures.'

How the Vote Played Out

The elected xGov Council found the proposal compliant with terms and conditions by a 6²¹ yes / 2 no / 3²² abstain vote, but its members split on substance. Councillor Simon Belingar called it 'a fairly new project with impressive social profiles and some early traction' and said he would want to see the stack used longer before judging its ecosystem value. Councillor Uroš Hudomalj questioned why a cUSD wrapper existed at all instead of using USDC directly, described the repo's PyTeal contracts as 'outdated tech' (Algorand's tooling guidance has since moved to Algorand Python as the successor), asked about Sui references in the codebase and the CONFIO token's role, and reminded the proposer that xGov is retroactive-only, making promises about future spending irrelevant to the ask. Fellow councillor Mohamed Ghiasi pushed back: 'A big yes from me' — '100% deserves the asked amount and even more IMHO.'

On-chain, the verdict was decided by turnout, not support. The proposal's app record shows 81 members voted against a 94-member quorum, with 471,133²³ approval weight against 4,499 rejection — short of the 317,545 weighted quorum threshold. The xGov portal accordingly shows 'Majority Approved' next to 'Proposal Rejected.' On March 25 the founder posted a vote drive: 'Our proposal already has strong majority support, but we are slightly short on quorum.'

The resubmission cleared the bar. Its on-chain app (3510819337, whose stored title decodes to '[Resubmit] Confío: Web2-to-Algorand Consumer Onboarding Infrastructure (LATAM)') drew 75 voters against a 60-voter quorum — 437,774 approval weight, 4,270 against — and the xGov platform's payout transaction on May 25 paid 250,000 ALGO to the proposer in the same atomic group that flipped the proposal's status to funded. The proposer forwarded 250,003 ALGO the same day to the treasury account that administers the cUSD contract. The thread itself discloses that a 7,500²⁴ ALGO proposal bond was posted using 'a portion of capital provided by early supporters.'

The Chain Move: Following Ondo to BNB

The migration is documented in the repo's BSC deployment record: the CusdPlusVault ERC-1967 proxy was deployed July 10 at 0x3C29417e…63Ed1²⁵, upgraded July 13 and July 20 through a 3-of-5 Safe, and source-verified on BscScan and Sourcify. Duende Limited — the seller named on the Confío App Store listing — was approved and whitelisted for Ondo Global Markets on July 30, and an Ondo Stocks router followed on August 10. Also on BSC now sit the $CONFIO token, presale, reward and vesting vaults, merchant-payment and payroll contracts, and an EIP-7702 sponsored-batch delegate that plays the role the Algorand atomic-group sponsorship played before.

The whitepaper²⁶ states the rationale plainly: 'The single-network design is not a generic chain bet. It follows the product's economic center of gravity: Ondo Finance made its USDY, InstantManager, price oracle, and USDT subscription/redemption path available on BNB Smart Chain.' Consolidating everything on one network avoids bridges and fragmented balances. The trade-off is named in the project's own risk table: BNB Smart Chain dependency — 'network interruption, validator coordination, congestion, gas-price changes, RPC failures, or ecosystem-level policy changes' — is listed as a residual risk. That is the other side of the ledger from the single-block finality and sub-cent fees the proposal originally sold to xGovs.

The token deserves its own scrutiny. Per the tokenomics document²⁷, $CONFIO is a fixed 1-billion BEP-20 with no minter, of which 893.6 million²⁸ — 89.36% — is the founder allocation on a 36²⁹-month vest, a concentration the document says is 'intentionally' shown as a founder allocation rather than an ecosystem reserve. The public presale sells 74 million³⁰ tokens on a continuous price curve from US$0.20 to US$1.30³¹, and the site's counter shows US$3,983 raised so far. An Algorand ASA named 'Confío' (unit CONFIO, 1-billion supply) also exists — its creator was first funded by the same account that created cUSD, so it is affiliated — but the tokenomics document designates the BSC contract as the canonical token.

What remains on Algorand is the cUSD contract, still live, and the open-source code. The whitepaper's next proof point is 'sustained consumer adoption on BNB Smart Chain,' and development is continuous: the repo's most recent commits, dated August 18, touch Koywe fiat-rail fixes and cUSD+ yield features, from a founder who alone accounts for 2,134 of the repository's contributions.

Where the Grant Went

The 250,003 ALGO paid on May 25 landed in the account that administers the cUSD contracts, which as of August 20 holds about 143,600 ALGO and 49 million of the Algorand-issued CONFIO tokens. Its on-chain history shows 158,194 ALGO moved out on August 15 and 12,348 USDC-a on August 20, both to the same address, which holds Tinyman's USDC-ALGO pool token and roughly 906,000 USDC-a. Meanwhile, the milestone report the proposal promised to publish to the xGov forum by month six does not appear there: a forum search for 'Confío' returns only the proposal thread itself, whose last post is dated April 29. None of this implies misuse — the transfers are simply the observable record of a grant requested as 'operational continuation' being spent.

What the Episode Says About xGov

This is a clean case study in what retroactive funding can and cannot buy. xGov paid for Algorand work that was real: open-source PyTeal contracts, a deployed non-custodial onboarding pipeline, sponsored-gas mechanics implemented at consumer scale, and thousands of wallets created through Web2 login — the exact traction figures in the proposal. What the program did not buy was ongoing chain commitment: roughly seven weeks after payout, the product's settlement layer moved to a network with a different validator set and finality model, a move the team documented openly rather than hiding.

The fair reading is not betrayal but reality: a lean two-person team pursuing the cheapest path to scale, which today means BSC-USDT liquidity and Ondo's permissioned rails. For Algorand, the retention question is what a grant actually keeps — the code stayed open source, the on-chain history stayed on Algorand, and cUSD still mints and burns on mainnet. But the milestone targets the proposal promised (150³² funded users by month two, 500³³ by month four, US$100,000³⁴ in cumulative USDC-a, 1,000³⁵ funded wallets by month six) were always going to be measured on whichever chain the product ended up using. xGov funds past work; it does not bind the future. Confío's arc makes that distinction concrete, and it is a distinction voters should carry into the next batch of 'Web2-to-Algorand' proposals.