Understand the receipt
The agent extracts the merchant, amount, currency, due date, and payment destination from an uploaded receipt.
reply.cash gives an AI agent the ability to understand a real-world bill, prepare an Avalanche-funded stablecoin payment, and monitor settlement—without giving the agent unilateral control of user funds.
Demo transactions use real networks and assets. Review every address and amount; start with a small amount.
Invoice detected
Agent-prepared · approval required
Fund with
Avalanche
Merchant receives
Solana
Route
Live quote
Control
User approval
Demo walkthrough
The demo separates agent reasoning from financial authorization, then preserves a status trail from quote to settlement.
The agent extracts the merchant, amount, currency, due date, and payment destination from an uploaded receipt.
It proposes Avalanche USDC or USDT as the funding asset and validates the destination, amount, and supported route.
The user reviews the asset, networks, recipient, quote, fees, and estimated time before any funds can move.
The payment state moves from quoted to deposit pending, cross-chain pending, and finally settled, failed, or refunded.
Agentic payment architecture
The agent handles interpretation, validation, route preparation, and status monitoring. The wallet owner retains the decisive action: reviewing and authorizing the transfer.
Turns unstructured payment context into a structured, reviewable intent.
The agent never receives a seed phrase and cannot bypass wallet authorization.
Uses Avalanche stablecoins as a funding rail for supported destination networks.
Carries amount, recipient, quote, timing, reference, and terminal outcome as one lifecycle.
Select an Avalanche funding asset, enter a destination, review the live quote, and observe the payment lifecycle. No transaction is sent until you explicitly approve it.