papers/morpheus/adoption/adopt-05-bank.tex

\documentclass[11pt,a4paper]{article}
\usepackage{../shared/luxcover}   % black cover (self-contained, TikZ logo, no assets)
\usepackage{../shared/proposal}   % \proposalhead helpers and the stake environment
\usepackage[utf8]{inputenc}\usepackage[T1]{fontenc}
\usepackage{amsmath,amssymb}
\usepackage{listings}\usepackage{xcolor}
\input{../shared/lstlang}

\newcommand{\code}[1]{\texttt{\small #1}}
\setlength{\emergencystretch}{3em}

\title{Accounts and Cards}
\author{Zach Kelling\\\textit{Lux Industries, Inc.}\\\texttt{[email protected]}}
\date{2026}

\begin{document}
\luxcoverpage
\sloppy

\proposalhead{Accounts and Cards}{%
A fiat rail the network runs: accounts, cards, and fiat in and out over MOR and stable balances. The
smallest dependency in the set. It needs nothing else here to be useful.}{ADOPT-05}

\noindent\textbf{The problem, in their numbers.}
MOR trades about \$57k in a day [verified]. DEX depth is about \$1.3M, most of it
protocol-owned in one Safe [verified]. A provider earning MOR has no fiat exit the network runs. To
turn MOR into rent or payroll they leave for an exchange and a bank Morpheus does not operate, and
they sell into that thin book. \code{api.mor.org} is OpenAI-compatible, so a Web2 customer buys
inference without a wallet [verified]; but that customer keeps no balance, holds no card, draws no
statement. The rail that carries inference stops at the wallet boundary. Every real customer crosses
it by hand.

\vspace{0.35ex}
\noindent\textbf{The fix.}
\code{luxfi/bank} is that edge, and it is built. One Go service, \code{bankd}, on Hanzo Base, with
auth, storage, and an admin console beneath it. It holds accounts, individual and business, with
multi-currency balances tracked as available and held. It issues virtual cards behind a KYC and
consent flow, quotes and executes fiat/crypto exchange, and takes deposits and sends. Fiat moves over
CurrencyCloud and IFX. KYC, AML screening above the reporting threshold, and sanctions and PEP checks
run as record-level hooks in the write path, not a layer bolted on after. Custody is three questions
kept separate (\code{COMPLIANCE.md}): who holds the signing key, where the asset sits, who may freeze
or reissue. A self-custody deployment leaves the customer holding their own key; bank records rather
than signs. Everything answers under \code{/v1/bank}, so the gateway gains a fiat rail without a
second API host.

\vspace{0.35ex}
\noindent\textbf{The outcome, quantified.}
A fiat on/off-ramp the network runs. Today MOR's only exit is a DEX clearing about \$57k a day
against about \$1.3M of mostly protocol-owned depth [verified]. A provider drawing more than pocket
change cannot cash out at scale. bank routes fiat over CurrencyCloud and IFX, independent of DEX
depth, so a provider's off-ramp stops being bounded by that book: draw MOR to a card or a bank
account, hold a stable balance, spend it. bank is also the smallest thing to adopt. It depends on
nothing else in the set, so it ships first and de-risks the rest, and it is the custody leg ADOPT-04
lends against.

\begin{stake}{Assumptions, stated}
\item Reach converts. The 100{,}000{,}000+ launch reach across Hanzo and Zoo surfaces [client] feeds
      the funnel that fills accounts [illustrative]. Not promised; the condition under which the rail
      fills.
\item The KYC base carries. Users onboarded through Hanzo IAM and \code{lux/wallet} [verified repos]
      enter bank's compliance path without a second onboarding.
\end{stake}

\noindent\textbf{Relevance.}
bank already exists, carries compliance in the write path, and its self-custody deployment means
Morpheus never holds a key. The Hanzo IAM and \code{lux/wallet} identity the funnel fills is the same
one bank onboards, so the fiat edge and the demand that makes it worth having land on one base.

\vspace{0.35ex}
\noindent\textbf{What this does not solve.}
An edge is not the settlement layer. bank holds and moves balances; it does not price inference,
prove it happened, or move MOR across chains. Which licenses apply in a given jurisdiction is a
question for counsel, not code. And a card is worth most once there is revenue to draw against it:
bank is useful the day it ships, most valuable once real inference revenue flows through the gateway
(ADOPT-07).

\vspace{0.5ex}
\noindent{\color{muted}\footnotesize\sffamily
ADOPT-05. Component \code{luxfi/bank}, live. The custody leg ADOPT-04 builds on; most valuable once
ADOPT-07 demand flows. \quad\textbullet\quad \texttt{[email protected]}}

\end{document}