NEXXEN

NEXXEN gives business AI agents a controlled way to pay for the tools they need. Owners set the limits, agents can act within them, and exceptions come back to a human. Our goal is useful agent spending without unrestricted financial authority. The demo follows an agent that needs compute to render product images before its client pays. A dashboard brings together spending rules, financing quotes, payment activity and human approval. Small requests proceed under deterministic policy. Larger requests are blocked, with a scoped approval path and an audit trail that preserves the original block. The working prototype demonstrates AP4M payments in RLUSD on XRPL EVM testnet, including a 0.001 RLUSD payout and settlement independently confirmed through public RPC. Separate earlier tests verified an AP4M receipt with the official SDK through Quicknode and a readable, one-use Ledger Flex approval for a 0.02 fixture RLUSD exception. That approval created a reservation, not a payment. Marqeta was tested separately with a $1 sandbox authorization hold. Financing quotes use fixtures; stablecoin-to-card funding and GPU rental/delivery remain simulated. Cash-flow credit is a future direction, not a live loan product. Any real debt would belong to the business. Built by Joshua Walter, Edwin Gutierrez and Jermaine Ulinwa.