Our Glass Redcurrant

Disclaimer. This is a concept note, not a finished system, and it says so plainly. Good things start small. Behind it sits a working sandbox. The ReAct loop runs in a test harness, and the first rule, planet-first (draft-planet-first.json, v0.3.0-draft), is implemented with tests. A few of its choices are still open. The next milestone is the first end-to-end payment on Solana devnet with its trace published on an audit page. When a payee wants it in a bank account, a regulated stablecoin off-ramp would carry it the last mile into SWIFT; that's a later step, not part of the sandbox. The capital pillar essays, the part of the ontology that defines terms like money, value and legitimacy, are in progress.

What's described here is the plain-vanilla base case: any ordinary payment that is made only when it's needed, and that carries its own record. More specific applications would build on it later. One example is distributing a basic income using the same rules (basic wants, planet-first, local-first, abstain), but that's a future direction, not something that exists today.

One Wants To Trust

On rails like Solana, stablecoin transfers settle in seconds for a fraction of a cent. The rails are close to free, and freer in that other sense of being more open to anyone; this brings us enormous opportunities but also entails risks.

But why enter a transaction? Capital, which is to say money, is under the best circumstances a token which flows freely in society and one whose issuance (quantity in circulation) matches our requirements: just enough for who need it to satisfy needs and on the whole keep people prosperous and content.

The great financial crisis however also showed that this ideal is best expressed when data on capital flow is transparent and relative creditworthiness (trust) is fully expressed, i.e. where disclosure on balance sheets and treasury is non-negotiable and verifiable with proof of reserves and trust is measured against consistent objective metrics.

If a penny saved is a penny earned and we maintain the standard that only essential purchases are made it's because we either don't have a surplus to spend or we understand the implications of consumption for the environment.

Ultimately only transactions having a zero impact on the quality of our air, water and soil must be permitted. If that is not a realistic objective today, then it must be a long-term one to which a commitment must be made; if for no other reason than because we are in the midst of a climate catastrophe - one which will continue unabated, entailing untold misery for everything that lives on this planet. Rather than hector people into changing the objective here is to make it materially beneficial by improving the quality of their lives and the quality of what they consume.

Stillness And Speed

Now that we have speed, the focus must be on what stillness (versus stagnation) entails; because while we need to use our money we also tend to preserve some for an uncertain future; the question is how much, because the stark and honest reality is that one person's comfortably balance in credit is another's anxiety-inducing debt; unless that is the capital is soundly allocated -which is why trust founded on transparency is so important.

Ideally we live our life day by day with as little lending, borrowing, and spending as necessary to make life comfortable; only that which is required and only that which allows for a sustainable draw on our common commode, the storehouse that is the Global Commons (our ecology). Money which moves quickly will ultimately enable this. It may even one day enable us to pay each other to improve our health, based on an objective measure of what is optimal, or to ameliorate our impact on our environment.

Argonaut's 'Our Glass Redcurrant' is a rules engine parsed from an ontology; a broad synthesis of practical concepts, accepted world models and metaphors. It is designed to help Large Language Models retrieve what matters when queried, and it leans on a mythology which anchors the model on an optimistic fact: The planet will always prevail and restore equilibrium on way or another, so it doesn't wander off into catastrophic hallucination. The physical planet will always find an equilibrium. Earth is a self-regulating, incredibly resilient system that has survived extreme cataclysms over its 4.5-billion-year history, including global ice ages ("Snowball Earth") and asteroid impacts that wiped out entire eras of life.

The principle at work is target fixation. Anyone who has learned to ride a motorcycle knows that if you stare at a truck's tires while passing at speed you will steer straight into them. Systems are no different; they drift toward whatever shaped their design (invariably whatever remains in frame). And the cultural urge today is to model and fix something external to ourselves personally. Argonaut focuses on the human journey rather than modifying a planet that is already made.

To navigate this, the system draws on a foundational worldview - a corpus Regnum Dei - which functions as a workflow sitting directly atop sovereign regulatory frameworks and structured financial standards like ISO 20022.

Generation

There are of course products that we absolutely need in order to live, and we may acquire them through transactions with counter-parties whose approach to creating them is regenerative. These are the sorts of small-scale enterprises that already in exist and which might now be allowed to flourish, but only if we're intentional about making quality a priority over quantity. The same can be said for both clothing and shelter; however chances are you really don't need either if you're in a position to read this.

Every transaction is a vote for the world we want in an increasingly global context. As digital payments evolve to close the gap between clearance and settlement (the space between you paying for something and somebody actually having the same money you used), so too is that world manifested ever more quickly. By taking responsibility for the choices we each make, and letting others do the same, we become direct participants in a wholesome form of democracy wherein we generate social justice personally governing our lives in way which diminishes over time the bifurcation between citizen and consumer.

Now, a transaction involves trade-offs. At a basic level we are exchanging money that we earned generating value for something we believe to be worth exchanging it for. For whatever reason, we feel that a transaction be afforded a degree of privacy; as things stand, however, it is not. Every single item on your receipt is parsed by various intermediaries (banks, credit card companies etc.) to paint a comprehensive picture of what goods and services are in demand at any given moment. How they use this data is not a matter for discussion, however. There is certainly no debate that the information, anonymized or otherwise, is enormously valuable to everyone who shapes institutional policy, public or private.

However, a topic of broad conversation today is how this data might be used to align and benefit interests: individual, sovereign and global (framed as ecology). This can be done by assessing how well each transaction serves all three at once. This data should be ours alone to make available to others if it's in our own personal interest to do so.

Our need to make better decisions on this score (trust) has never been more important because of the speed with which money (or for that matter any form of data) moves and the risk this entails, simply because money enables power and authority - often over others and their environment. We now have the capacity to assess trust at the most granular level possible - identity. Those jurisdictions whose enterprises and communities embrace these options will thrive because they enable autonomy, agency and productivity, which result in both prosperity and contentment.

A world that is not only prosperous but one in which people are creative and productive requires us to think of money as the shared tokenized representation (or abstraction) of the things we need to live a good life, and which flows to enable it. The question then is how many such tokens do we need in circulation to ensure everyone has the capacity to use them to access basic needs (potable water, shelter, food, clothing), so that they don't feel the need to hoard capital as a contingency for uncertainty?

It's a difficult question - but one made easier with transparency. The more transparent we are individually, the more easily others may come to a decision as to whether to trust us, and the greater the sense of collective security that may be afforded.

Divine Governance: Our Glass Redcurrant

The workflow is called Our Glass Redcurrant. The currant takes its name from Corinth, which sat on the isthmus between Athens and Sparta as a node where transactions could be mediated between the great rivals. Currency and current share a different root, the Latin currere, "to run": value that flows. A current has long been considered an aspect of the divine in the body (qi, prana, the Holy Spirit). The synchronicity of terms expressing flow in different contexts is based in the fact that all capital comes from ideas, which are then actualized to generate value. Wine, as well of course, remains for many a significant store of value.

The workflow starts with a need, expressed as intent: pay X for Y. That intent arrives as an ISO 20022 request (a payment initiation or a request-to-pay) carrying the payee, the amount, a purpose code, remittance details and often a location. Argo reads those fields first, and they become the factors the ontology and rules weigh. Argo, the agent at the heart of Argonaut Agentic, takes that intent into a ReAct loop. It reasons, acts, observes and reasons again. The reasoning is performed by a language model, currently Mistral Small. The model interprets the intent and proposes options without making a decision. Each option is tested in code against a rules engine: versioned JSON rules written from the ontology and applied by a fixed engine. That ontology is the philosophy published on this site as an auditable reference, much as those in so-called private and public governance are called to account in parliament, corporate town halls and quarterly earnings calls. The ontology functions as an identity embodied as a utility. It transforms truth, defined as the factors of a transaction between counterparties, into a meaningful outcome: an answer as to whether the transaction should be processed because it aligns the interests of the individual with those of the relevant sovereign and the ecology.

The ontology works on two levels. Canonical terms give the agent a fixed vocabulary, so words like transaction or trade mean the same thing every time they're used. The essays elaborate the nuances, explaining complex notions such as intelligence, identity and capital itself through metaphor, since each is ultimately an aspect of human behaviour. The ontology is what the agent reasons with.

The rules layer is the part that actually decides. The rules are written from the ontology, but they aren't the ontology itself. They're fixed and versioned:

• Abstain first. "No transaction" is the starting point, at zero cost. A payment has to earn its way past it by meeting a real need. Sometimes the best payment is the one you don't make.
• Planet-first. Options are weighed by their impact on air, then water, then soil, in that order. Some options improve them: buying from a regenerative farm counts in the planet's favour.
• Smallest-first. The least extractive option that meets the need comes first.
• Agency. The payer's own judgement stays in the loop. The agent advises and records; it doesn't override. Nobody is governed on your behalf.
• Local-first. Options are queued in rings outward from the payer, so nearby suppliers are considered before distant ones.

Out of that comes a decision. If the need doesn't outweigh the cost to the planet, the answer is abstain: Argo defers the request back into the queue, so nothing moves and nothing is spent for now, until the need outweighs the cost or a better local option turns up. The engine prioritizes; it doesn't gate. If it's pay, Argo composes the outgoing ISO 20022 message with purpose and remittance codes, the same standard the banking world has been steadily adopting, and settles it in USDC on Solana, where the funds land with the payee in seconds rather than in three to five business days.

Alongside the payment, Argo writes a reasoning trace: which rules applied, which options were weighed, and how the payment cleared the default of not paying, or what held it back. That trace is hashed, and the fingerprint goes into the transaction memo on-chain. The full trace is published on a public audit page at transition-insight.org. Anyone can take the trace, hash it and check it against the chain. If it matches, that is the reasoning the money moved on. The fingerprint on-chain is permanent, which is a quiet kind of confidence.

The transaction register

In Lens & Narrative the transaction is a register of its own. Is a transaction balanced and minimalist, or noisy and polluting? Is it decadent, or sensitive to those who go without? Those questions normally get asked afterwards, when there's time, and by people doing their best with what's left on the record. Argo asks them before the money moves and keeps the answer.

That means the rules have a bias, and they say so. Bias isn't bad, and alignment isn't neutral. Every payment system already encodes preferences, speed and convenience among them, and they serve people well; they're simply left unsaid. Argo's preferences are written down, versioned and open to challenge. Abstain-first is a bias. Planet-first is a bias. Local-first is a bias. You can disagree with them, fork them or weight them differently, but you can see them.

Open on purpose

The ontology and the rules are open, and that's deliberate. I want these ideas copied. If someone takes the planet-first rule and builds a better agent around it before I do, good. The point is the transition, not the patent.

Where the value lies

Open doesn't mean there's nothing worth paying for. The value isn't in moving money; the rails already do that for almost nothing. It's in what surrounds the movement:

• Audit reports. Co-ops, charities and small organizations with ESG or grant reporting obligations spend real time accounting for where money went. Argo's traces are that account, produced as the money moves rather than reconstructed afterwards. Reporting becomes a by-product, and that time goes back to the work itself.
• Credibility for good vendors. Suppliers whose work improves air, water or soil, such as regenerative farms, gain a public, checkable record that they were chosen for it. That visibility is worth something to them, and it's earned by the work itself.
• A hosted agent. Anyone can run the open version. Plenty of organizations would rather not.
• Fitting the rules. Every organization has its own policies. Translating them into a rules layer the agent can apply, and audit, is real work.
• Public-goods funding. Work that is meant to be copied is a natural fit for grants that support open infrastructure.

Design choices

These are the choices the design is built around. They're how it's meant to work, not a description of finished code.

Privacy. The public trace carries no personal data. It holds the rule IDs and versions, the options weighed and their scores, and the decision. Payer and payee appear only as pseudonymous addresses. Anything private stays off the page. Public front end, closed back door: the decision is open and your data stays yours.

Who decides. A language model (any one will do) proposes options and explains them. The rules, written from the ontology, are fixed, deterministic, versioned code with tests, and they score or gate those options. The same inputs give the same verdict, and any conflict is settled by the written rule, not the model's mood or anyone's discretion. Whether the planet check gates options outright or only ranks them is still one of the open choices.

Order of operations. The trace is written and hashed first. Then the payment is submitted with that hash in its memo. Then the trace is published. Because the hash verifies itself, a late or failed publish delays the audit page but can't corrupt it. Anyone holding the trace can check it against the chain.