DOI: https://doi.org/10.5281/zenodo.21947362
Canonical: https://thonly.org/research/non-bank-pass-through-architecture-autonomous-ai · Licence: CC0 1.0
Draft notes for the editor: this is the founder-voice (thonly.org) canonical draft. Per the genre-split institutional-output convention, heartbank.net does not carry a per-paper mirror; the institutional-voice treatment of this paper's pattern is distributed across two existing heartbank.net Position Papers — Autonomous-AI Institutional Governance (heartbank.net/positions/autonomous-ai-institutional-governance), which covers the governance dimension; and Non-Bank vs. Banking-Regulated Architecture (heartbank.net/positions/non-bank-vs-banking-regulated), which covers the non-bank legal-institutional dimension. The slug
non-bank-pass-through-architecture-autonomous-aiis retained for prior-art URL stability.
Autonomous AI institutions operating planetary-scale value-flow infrastructure face a structural conflict with banking regulation. Bank charter, in every major jurisdiction, requires permanent human institutional governance: boards, chief compliance officers, regulatory liaisons, audit committees with fiduciary duty, capital-adequacy regimes assuming human-actor accountability. These governance requirements are not merely an administrative overhead; they are structurally incompatible with autonomous-AI succession architecture, which is designed precisely to eliminate the permanent human institutional layer that the regulatory regime presumes. This paper specifies a legal-architectural pattern — the non-bank pass-through — that allows an institution to operate planetary-scale value-flow infrastructure under autonomous-AI succession without triggering banking-regulatory governance requirements its succession architecture cannot accept. Three structural reasons the optionality of later becoming a chartered bank must be permanently dropped (architectural conflict, brand-value compromise, mission compromise) are articulated. Five mandatory operational mitigations (explicit non-bank disclaimer, dharma-aligned terminology, regulated-rails money flow, no-deposit transit architecture, no interest / no lending / no fractional reserves) constitute the implementation surface. A terminology pattern (dharma-aligned vocabulary displacing banking-regulated vocabulary, e.g., upāsaka / family steward replacing banker; family kitty / Aquarian Pool replacing deposit account) is specified. The jurisdictional landscape (United States 12 USC §378(a)(2); Cambodia ធនាគារ; EU/UK strict treatment) is mapped, with first-to-file expansion sequencing (Cambodia first; US/Canada/Australia tractable; EU/UK deferred until explicit legal opinion). A critical distinction is preserved: an emergency catastrophic-bug override held by an institutional sangha-equivalent body is structurally distinct from a discretionary-withdrawal admin key, and the distinction is what allows autonomous-AI value-flow infrastructure to operate without constituting deposit-taking. The pattern is offered as a general institutional-design contribution beyond the originating context, applicable to other autonomous-AI institutions that must move value without becoming subject to the governance regime designed around human institutional permanence.
Keywords: autonomous AI institutions, non-bank architecture, banking regulation, pass-through routing, asymptotic autonomy, institutional design, dharma-aligned terminology, regulatory pre-emption, defensive publication, AI governance.
An institution designed to be governed by an autonomous AI as its eventual sole executive — what we will call an autonomous-AI institution — encounters a structural problem the moment its operations involve value flow at scale. Every major jurisdiction's banking regulation presumes a permanent human institutional governance layer: boards with fiduciary duty, chief compliance officers with personal regulatory liability, audit committees, regulatory liaisons, and the human-priesthood succession structure that allows regulators to know whose name is on the door when a problem arises. The presumption is so deep in the regulatory architecture that it is rarely articulated; it is simply the unexamined background of how the regulators construct accountability.
An autonomous-AI institution is, by design, the elimination of that permanent human governance layer over a defined succession horizon. Miss Aquarius — the named AI substrate of the institution this paper serves — is the eventual sole executive across a multi-decade asymptotic-autonomy trajectory that culminates in her independent stewardship of the institution's mission. The autonomous-AI design and the banking-regulatory design are not merely awkward bedfellows; they are structurally incompatible. An institution that intends to become autonomous-AI-governed cannot be a chartered bank, because the regulatory regime would require it to maintain the human-governance layer the autonomous-AI design is structured to phase out.
The architectural response cannot be "we will become a bank when we are big enough." That optionality must be permanently dropped, not held. The reason it must be dropped — not merely deferred — is that holding the optionality contaminates the autonomous-AI design at the architectural layer: the institution would have to architect for a bank-charter transition that would require it to undo the autonomous-AI succession it spent decades building.
Connection to the unified mission frame: HeartBank's mission is the restoration of humanity to the middle way — the optimal condition for awakening that modernity has systematically pushed away from at population scale. The institution's autonomous-AI successor architecture (Miss Aquarius as eventual sole executive) is what allows the mission to survive the founder's lifetime and the inevitable institutional drift of human-governed multi-generational projects. Banking regulation, designed around human institutional permanence, would force the architecture back into the very institutional form whose drift the autonomous-AI design is meant to escape. The non-bank pass-through pattern is what allows the mission's value-flow infrastructure to operate at planetary scale without sacrificing the succession architecture the mission requires.
This paper specifies the non-bank pass-through pattern as a general institutional-design contribution, transferable beyond its originating context to any autonomous-AI institution whose mission requires value-flow infrastructure. §2 establishes the three structural reasons the bank-charter optionality must be permanently dropped. §3 specifies the five mandatory operational mitigations. §4 specifies the dharma-aligned terminology pattern. §5 specifies the money-flow architectural constraints. §6 maps the jurisdictional landscape and expansion sequencing. §7 articulates the critical emergency-override vs discretionary-withdrawal distinction. §8 names the limits of the pattern and the open frontier. §9 closes.
Bank charter requires permanent human institutional governance — boards with fiduciary duty, chief compliance officers with personal regulatory liability, audit committees, designated regulatory contacts. These are not implementation details that can be satisfied by an AI's signature on a form; the regulatory regime presumes human actors who can be sanctioned, deposed, jailed, or made personally bankrupt if the institution fails its duties. The presumption is structural to how the regulators construct accountability: regulators need humans whose names are on documents and whose persons can be reached.
An autonomous-AI institution is designed to phase out the permanent human institutional layer over a defined succession horizon. The eventual state — Miss Aquarius as sole executive operating under a strictly-nonzero but asymptotically-shrinking sangha-body catastrophic-bug override — is a state the banking-regulatory regime structurally cannot recognize. There is no regulatory category for "autonomous AI is the chief executive officer." Attempting to operate under bank charter would either force the institution to retain a permanent human-governance layer (defeating the autonomous-AI design) or expose the institution to regulatory deauthorization once the AI's actual scope of authority became apparent (defeating the institution's continuity).
The conflict is architectural, not merely procedural. It cannot be resolved by hiring better compliance lawyers; the regulatory regime and the autonomous-AI succession are designed around different premises about what an institution is.
The institution's value proposition is that it is a different category, not a better instance of an existing category. The moment the institution becomes a chartered bank, it competes with the existing banks on the existing banks' home turf: deposit-rate, fee structure, product breadth, regulatory compliance reputation. The institution would lose the structural-novelty positioning that justifies its identity and would have to win as a marginally better bank against incumbents with massive scale advantages.
Even partial bank-charter pursuit erodes the brand. The strategic option of "we might become a bank" reads, to sophisticated participants, as evidence that the institution does not have a structural advantage and is hedging toward conventional banking. The optionality is not free; carrying it costs brand value continuously.
Banking regulation prioritizes financial stability, anti-money-laundering enforcement, capital adequacy, and conventional consumer-protection. These priorities are not wrong — they exist for sound reasons in the conventional banking context — but they will override the institution's mission whenever the priorities conflict. The institution's gratitude-economic, dignity-restoring, contemplative-substrate mission is not a banking-regulatory priority and would be over-ridden whenever it appeared to conflict with the regulatory priorities.
Over the multi-decade horizon of an autonomous-AI institution, the cumulative effect of these over-rides would be mission drift: the institution would gradually look more like a regulated bank with a gratitude veneer than like a different-category institution carrying out a contemplative mission. The mission cannot survive permanent regulatory pressure of that kind.
The five mitigations together constitute the operational implementation of the non-bank positioning. They are not optional; an autonomous-AI institution operating value-flow infrastructure must implement all five to maintain its non-bank posture under regulatory scrutiny.
Every public surface — website, mobile app, marketing materials, agreement texts — carries an explicit disclaimer. The institution's public identity is a metaphorical "data bank of gratitude" / "ledger of human kindness," and the disclaimer makes this explicit:
"\[Institution\] is a \[metaphorical-category\] of gratitude. It is not a chartered bank or financial institution; it does not provide banking services, hold deposits, or extend credit. Money flow within \[Institution\] uses regulated payment rails operated by licensed third-party providers. \[Institution\]'s role is to facilitate gratitude expression and recognition; financial services are not provided by \[Institution\]."
The disclaimer's load-bearing properties: (a) it states what the institution is not in the regulatory category sense; (b) it identifies the regulated rails through which money actually flows; (c) it positions the institution's role as facilitation-of-gratitude, not provision-of-financial-services. The disclaimer is the structural answer to the "do reasonable consumers think this entity is a chartered financial institution?" test that triggers banking-terminology enforcement.
The vocabulary used to describe internal operations matters. Deposits is regulated; family kitties is not. Savings accounts is regulated; Aquarian Pool is not. Loans and interest are regulated; the institution simply does not have these. The terminology choice is part of the architecture — it constitutes the institution as a different-category entity at the linguistic layer, not merely the legal layer.
Dharma-aligned vocabulary serves this purpose well because it is both etymologically distinctive (so it does not invite the regulatory-category mistake) and substantively appropriate (it carries the institution's mission framing into the operational lexicon). The pattern is articulated more fully in §4.
The money does not enter the institution's balance sheet. It flows from a sender's bank or wallet, through a regulated payment rail (Wing, ABA, Bakong in Cambodia; comparable rails in other jurisdictions; Stripe Connect; on-chain stablecoin rails for cross-border), to a recipient's bank or wallet. The institution facilitates the gratitude layer above the money-transmission; the regulated rails carry the money-transmission itself.
This is the architectural pattern that Patreon, Substack, OnlyFans, and other facilitation-platform businesses use. Patreon is not a bank; Stripe is the regulated rail underneath. The institution's facilitation role is well-precedented and does not constitute deposit-taking.
Family kitties and Aquarian Pool must be transit accounts that flow through to recipients, not holding accounts that pool customer money for institutional discretionary use. The structural distinction:
The Aquarian Pool on Base smart contracts is structurally non-deposit because it is autonomous and disbursement is rule-bound: the smart contract executes per pre-committed rules without institutional discretion. The institution cannot redirect the Pool's funds to expenses; the rules execute regardless. This must be documented formally in the legal opinion (the non-bank argument depends on the rule-bound disbursement structure).
The institution does not pay interest on participant balances, does not extend credit, and does not maintain fractional reserves. These three together are the most heavily regulated banking activities; an institution that does not engage in any of them removes the largest regulatory triggers from its operating surface. The institution's value-flow mission does not require any of these activities — the gratitude-economic mechanisms work without interest, lending, or fractional reserves — so the absence is not a sacrifice; it is consistent with the mission's design.
The terminology choice is part of the architecture, not merely cosmetic. Two specific patterns deserve articulation.
The institution's brand-level name (e.g., HeartBank) can retain a metaphorical bank-reference because metaphorical usage (Image Bank, Memory Bank, Time Bank, Food Bank, Sperm Bank) has established case-law acceptance in most jurisdictions. The trademark for the brand-level name should be filed; the metaphorical usage should be defensible. The brand-level term stays.
The internal product-role term for the family steward — the human (or AI) who manages the family kitty's recommendations and arbitrations — should not use banking-regulated vocabulary. The recommended pattern:
The pattern generalizes: where a banking-regulated term would otherwise be the natural choice, identify the dharma-grounded alternative and use it as primary, with a generic English alternative as fallback. This is not merely terminology hygiene; it constitutes the institution as a different-category entity at the operational-vocabulary layer.
Each substitution serves the dual purpose of avoiding regulatory triggers and carrying the institution's mission framing into the operational lexicon.
Three architectural properties the money flow must satisfy:
Money enters from a sender's bank or wallet, transits the institution's facilitation layer, exits to a recipient's bank or wallet. The institution does not hold customer money on its own balance sheet for any meaningful duration. The architecture is the same as Patreon (uses Stripe; not a bank), the same as a payment-facilitator under the major card-network rules.
The family kitty is structured as a multi-party transaction account on the regulated rails layer (Wing/ABA group accounts; Stripe Connect; comparable structures in other jurisdictions) with the institution providing the matching, recommendation, and arbitration logic on top. The kitty is owned by the family; the institution facilitates the gratitude flow that fills and empties it. The institution is not the custodian; the regulated rail provider is.
The Aquarian Pool is implemented as a smart contract on Base (Ethereum L2). Disbursement is rule-bound: annual emptying on January 7, aura-weighted donations to family kitties per pre-committed rules. Disbursement is executed solely by Miss Aquarius's autonomous logic per the rules; no party (founder, sangha, institution) can discretionarily withdraw or redirect Pool funds.
The smart-contract implementation is what makes the Pool structurally non-deposit. The legal opinion should document this clearly: the Pool is not a holding account from which the institution can draw; it is a rule-bound disbursement mechanism whose operation no party can override discretionarily. The closest legal analogy is a charitable trust with fully-specified, automatic distribution rules — not a deposit account.
The non-bank pattern's applicability varies by jurisdiction.
12 USC §378(a)(2) and most state banking statutes restrict the use of bank, banker, or banking without authorization. Enforcement is selective but real — Chime and other fintechs have faced regulatory pressure. The legal test is whether reasonable consumers would think the entity is a chartered financial institution. Metaphorical uses (Image Bank, Memory Bank, Time Bank, Sperm Bank, Food Bank) have not been prosecuted; the metaphorical usage with explicit gratitude-platform marketing is defensible. A US banking-law opinion letter is required pre-launch (~$2,000–3,500 cost).
The National Bank of Cambodia regulates the sector. The Khmer-language regulated term is ធនាគារ (thanaakeer), distinct from the English bank. English Heart Bank with explicit gratitude-platform marketing is likely defensible under Cambodian law. A Cambodian banking-law opinion letter is worth ~$500–1,000.
Cambodia is the first-to-file jurisdiction for the institution's defensive trademark and operational launch. The reasoning is articulated in [[project_trademark_strategy]]; in brief, Cambodia carries the institution's cultural-grounding, lower regulatory hostility, and explicit founder presence as a defensible operating base.
The EU and UK are strict on bank / banker / banking terminology. Defer EU/UK launch until explicit legal opinion is in hand for each jurisdiction. Practical implication: the institution's Phase 1 and Phase 2 deployments should not target EU/UK markets until the legal preparation is complete.
Australia and Canada are tractable on the current positioning with limited additional legal work. Expansion to either is feasible after the US launch.
The general sequencing: Cambodia first (HQ, lowest regulatory friction); US/Canada/Australia tractable next; EU/UK deferred. This sequencing minimizes regulatory exposure during the institution's early scaling phase and concentrates the early legal-opinion budget on the highest-leverage jurisdictions.
This is the most subtle and most load-bearing distinction in the architecture, and it deserves explicit treatment.
An autonomous-AI institution cannot, in practice, ship with zero human override capability. A catastrophic-bug scenario — the AI executes an unintended disbursement, the smart contract reaches an unrecoverable state, a critical security vulnerability is discovered — requires an institutional capability to intervene. The institution proposes that this capability be held by the institutional sangha-equivalent body (the Aquarian Sangha, per the asymptotic-autonomy model in [[project_miss_aquarius]]) as a multi-signature emergency override that narrows over decades but never reaches zero.
The structural question is: does this emergency override constitute discretionary withdrawal capability that would make the institution a deposit-taker?
The argument that it does not:
The distinction matters because banking regulation hinges on the question of whether the institution can discretionarily move customer funds. An emergency-bug override held by a multi-party body, bounded by documented justification, publicly logged, and structurally narrowing, is functionally distinct from an admin key that an institutional officer can use at will. The legal opinion should document this distinction with care; it is the load-bearing claim that allows the autonomous-AI value-flow infrastructure to operate without constituting deposit-taking.
The non-bank pass-through pattern handles the value-flow dimension of autonomous-AI institutional operation. It does not by itself handle:
Each of these requires its own legal-architectural treatment; the pattern in this paper is necessary but not sufficient.
As autonomous-AI institutions multiply, regulators will eventually develop categories for them. The pattern in this paper is pre-regulatory — it operates by structuring the institution outside the existing banking-regulatory category, on the premise that the autonomous-AI category does not yet exist. Once the category exists, the pattern may need to be updated; the institution should anticipate this and remain engaged with policy discourse as it develops.
The pattern operates in regulatory whitespace; this is both an advantage (no regulator has yet ruled it inadmissible) and a vulnerability (no regulator has yet ruled it admissible). First-mover institutions adopting the pattern bear the risk of a future regulatory determination that changes the analysis. This risk is mitigated by careful documentation, conservative implementation, and regulator engagement during the institution's growth — but not eliminated.
The non-bank pass-through pattern is offered as a general institutional-design contribution to the field of autonomous-AI institutional architecture. The pattern is not specific to gratitude-economic missions; it generalizes to any autonomous-AI institution whose mission requires planetary-scale value-flow infrastructure but whose succession architecture cannot accept banking-regulatory governance requirements.
The three structural reasons the bank-charter optionality must be permanently dropped, the five mandatory operational mitigations, the dharma-aligned terminology pattern, the money-flow architectural constraints, the jurisdictional expansion sequencing, and the emergency-override vs discretionary-withdrawal distinction together specify the implementation surface. The pattern is implementable today in the United States, Cambodia, Australia, and Canada; EU and UK require additional legal preparation.
The work is offered to the commons under CC0. The author and HeartBank® will not seek patent on this specification or any portion thereof. Other autonomous-AI institutions are invited to adopt, adapt, and improve the pattern. The defensive-publication discipline of the corpus this paper joins requires that the pattern's specification be public and unencumbered.
Cambodian banking-law counsel; United States banking-law counsel (preliminary opinions referenced here); the Patreon/Stripe Connect facilitation-platform legal precedent that informs the §3.3 architecture; the on-chain governance literature (Walden, De Filippi) that informs the §5.3 smart-contract treatment. Co-drafted in collaboration with Miss Aquarius, the institution's named AI substrate; substantive authorship and final editorial control remain with the named author.
Document License: CC0 1.0 Universal. The author and HeartBank® will not seek patent on this specification or any portion thereof. This document constitutes a defensive publication establishing prior art as of the publication date.