The Problem Money Solves

Imagine you're a baker with bread to spare. Your neighbour is a carpenter who needs bread. Simple: you trade. But what if the carpenter doesn't need bread? What if she needs accounting help?

You'd need to find someone who needs bread, has accounting skills, and is willing to help the carpenter. That's three conditions, and they all have to line up at once.

The matching problem

Economists call this the "double coincidence of wants." As economies grew more complex, finding these matches became impossible for humans to manage.

Money solved this elegantly. The baker sells bread to anyone for money, then uses that money to buy anything else. No need to find the perfect match.

The cost

But money isn't free. It comes with:

  • Intermediaries who extract value (banks, payment processors, platforms)
  • Interest rates that make cash-strapped businesses choose between paying debts and investing in growth
  • Timing gaps: you need cash now to deliver value that won't be paid for later
  • Artificial scarcity: when credit tightens, businesses with plenty of capacity sit idle simply because the medium of exchange has dried up

For 5,000 years, these costs were unavoidable. There was no better solution.

What's Now Possible

The insight

A computer can check millions of possible exchanges in seconds. It can find paths that humans would never spot.

The baker doesn't need to find one perfect match. An algorithm can find a chain:

Baker gives bread to the graphic designer.
Designer does branding for the law firm.
Lawyer reviews contracts for the carpenter.
Carpenter builds shelves for the baker.

Everyone gives what they have spare. Everyone gets what they need. The loop closes without anyone needing cash.

Why now?

Finding cycles in a network of offers and needs is a solved problem in computer science. What's new is:

  • LLMs can translate "I've got Friday afternoons free and some old office furniture" into structured capability descriptions that an algorithm can match
  • Networks can form without centralised platforms controlling access
  • Trust systems can scale through track records, vouching, and graduated exposure

The technology to solve matching directly, without money as intermediary, now exists.

Key Principles

Surplus, not sacrifice

This isn't about giving away things you need. It's about giving away things that would otherwise go to waste:

  • The consultant's empty Friday
  • The theatre's dark afternoons
  • The designer's gap between projects
  • The caterer's unsold portions

If the alternative is "nothing," then almost anything is a better deal. That's the surplus baseline.

Subjective value

You decide what something is worth to you. No universal currency, no conversion rates.

When you help someone, they record how much that helped them. When someone helps you, you record what it was worth to you. Each participant keeps their own ledger.

This eliminates the endless arguments about whether an hour of legal advice equals an hour of cleaning.

Multi-party chains

Direct swaps are rare. Most value flows through chains of three, four, five, or more participants.

The algorithm finds these chains. You just describe what you have and what you need.

Protocol, not platform

SEP is designed as an open protocol, not a proprietary platform. The specification is open; implementations can compete. If any single operator fails or turns hostile, participants can move. This matters because single points of control create single points of capture.

Trust through relationships

Trust is built from verifiable identity, exchange history, and network position, not ratings or scores. Network position is expensive to fake because it requires actual exchanges with real partners over time, making gaming costly without reintroducing currency-like metrics.

These principles, along with others on accountability, governance, transparency, and sustainability, are explored in depth on our philosophy page, which also names the tensions we're actively navigating.

This Has Worked Before

90 years of proof

This isn't speculation. Mutual credit systems have operated at scale for nearly a century.

60,000 Swiss businesses
90 Years operating
CHF 2B Annual transactions

WIR Bank (Switzerland, 1934–present) operates as a fully licensed bank. Research shows it acts as an economic stabiliser: activity increases when conventional credit tightens.

Sardex (Sardinia, 2009–present) has 4,000 businesses with EUR 50+ million annual volume. True mutual credit with active trade brokers who know members and facilitate matches.

Both systems survived because of professional management, active matching, business focus, and accountability.

What previous systems couldn't do

WIR has density: 90 years of accumulated membership in Swiss SMEs. Sardex has brokers: humans who know every member's business.

Neither could match at speed across thousands of participants finding connections that span industries and geographies.

AI changes that. The algorithm can see the whole network at once.

What This Isn't

What people think

This is cryptocurrency. Another speculative token or blockchain project.

What it actually is

No token, no blockchain, no speculative asset. Balances aren't transferable. This is about matching, not money.

What people think

It's just bartering. Trading goods directly like in ancient times.

What it actually is

Multi-party chains, not direct swaps. You help someone who helps someone who helps you. The algorithm closes the loop.

What people think

An ideological critique of capitalism or manifesto for a new economic system.

What it actually is

A pragmatic exploration: could AI solve the coordination problem differently? If it works, it's a tool. If not, we'll learn why.

Where This Is

This is a proof of concept: working code that demonstrates the core mechanism.

What exists

Matching algorithm that finds multi-party cycles
Trust system with graduated exposure
Schemas for describing offerings, needs, and exchanges
Example scenarios showing the mechanism in action
Detailed design positions on accountability, trust, governance, and more

What doesn't exist yet

A live network (this is a demonstration, not a service)
Agent integration (human-only for now, agent participation is future work)
Deployment infrastructure (no servers running, just code you can explore)

We've analysed most of the hard questions in depth: accountability, trust, governance, bad actors, transparency. Some remain genuinely open. See our hard questions page for where we've landed.