About
A closed-loop control system for goodwill.
The mark you see on the home page is not a brand. It is a P&ID — a process and instrumentation diagram — and it is what we sell. Five shapes form a question mark because the system is a question. It never settles. It keeps asking, measuring, adjusting.
Hopper sets the heading. Filter is the white space between the shapes — what passes from one station to the next. Pump drives the work. Gauge measures reality, not feeling. Controller reads the gauge against the heading and asks the question: is this where we should be? If not, it adjusts.
The picture
It is a control system, not a metaphor.
Engineers draw factories this way before a single pipe gets welded. The same diagram describes a mind, a business, a venture, a life. The shapes are real. The loop is real.
The same inversion is winning in biology
Organisms harness stochasticity to evolve purpose at the level of the whole — not by deterministic gene blueprint, but by constrained autocatalytic process.
That sentence is Denis Noble's life work. He spent fifty years inside biology proving that the gene-centric reductionism we were all taught in school is falsifying — and that what replaces it is a system-level control loop, with constraints (boundary conditions, cell membranes, organ structure) doing causal work that bottom-up molecular description cannot see. Polygenic risk scores — the orthodoxy's promise to predict disease from genes alone — have now failed the drug-trial test in cardiovascular, cancer, and Alzheimer's research. Sixty percent of global deaths. ROI on cures from the genome project, near zero.
Function over molecule. Loop over part. Whole gives purpose to the parts. The same inversion the rest of this page argues for AI and commerce is the inversion biology is now being forced to make. The biology of the loop.
The thinking protocol it produces.
In control engineering, a negative feedback loop is what keeps a system alive. The word sounds wrong if you trained on social-media language — but it is the inversion that matters. Negative feedback converges. It pulls the system back toward a stated setpoint when drift starts. Body temperature. Thermostat. Auto-pilot. Every system that survives is a negative-feedback loop with a setpoint.
Positive feedback is the opposite. No setpoint. Whichever way the system tips, it tips harder. Engagement amplifies engagement. Outrage amplifies outrage. Spend amplifies spend. The internet, as currently shaped, is one giant positive-feedback amplifier. AI bolted on top makes it faster, not better.
For readers who prefer the voyage frame: the captain sets a heading. The instruments read the wind and the current. The controller asks whether the ship is still on the heading and, if not, turns the wheel. Same picture. Different vocabulary.
The problem
Most AI has no setpoint.
It accelerates whatever direction your loop was already pointed. If the heading was wrong, the drift is faster.
Most businesses that do have a setpoint have set it to profit. Profit is a token. It is the easy thing to measure. A loop with a profit setpoint converges toward extraction — even when extraction destroys the real value the tokens were meant to represent. Trust collapses. Spirit empties. The numbers go up while the substrate rots.
Most AI platforms are profit-setpoint loops on positive feedback. They sell you faster extraction and call it productivity. They do not put a controller back in charge. They make the runaway smoother.
The setpoint
Goodwill, not profit.
Profit is a tracking signal — like temperature for a body. The setpoint is the thing the token is supposed to track. Most loops have confused the two.
We use the ledger that already runs in every body, team, and market — six entries the gauge actually measures: goodwill (generosity without expectation; trust that survives storms), spirit (joy given freely; energy that does not deplete), flow (attention fully committed; mastery that earns respect), alignment (actions that match stated values; credibility that compounds), loyalty (commitment held through difficulty; relationships that endure), and love (orientation toward the good of others; a ledger worth keeping).
Goodwill is the highest-durability capital. It survives crises that wipe out the others. A token backed by a poisoned book is a memecoin. A token backed by a healthy book is a currency. Read the ledger.
Will + intent + attention
Willpower holds the heading.
The setpoint cannot be set by the machine. It is held by a human discipline against pressure to drift toward whatever the spreadsheet rewards.
Willpower is not effort. It is will plus intent plus attention — three things stacked. Will picks the heading. Intent commits to it. Attention notices when the ship has drifted before the gauge says so. Without willpower, the controller has no truth to act on. With it, AI becomes velocity inside the loop instead of direction over it.
The captain is still required. AI does not replace the captain. AI is the crew that the captain has finally been allowed to deploy at machine pace.
The purpose
Conscious evolution.
The loop does not just converge. It raises the level it converges toward — round after round.
Every cycle the controller asks the question. Every cycle the gauge tells the truth. Every cycle a small correction lands. Over many cycles the system evolves — not because anyone forced it, but because the loop is honest and the setpoint serves beyond self.
We call this a
Validated Virtuous Feedback Loop
. Validated, because the gauge reads reality. Virtuous, because the setpoint is goodwill. Feedback loop, because that is the only architecture that produces convergence toward flourishing instead of runaway toward extraction.
Why now
Commerce is rewiring around loops.
The protocols being laid this decade are the standard fittings between loops. Loops that trade with loops will compound. Loops with no setpoint will become invisible to the trading layer.
Buyers are starting to send their crew ahead. A machine reads your shop, checks your manifest, verifies your flag — and decides, in milliseconds, whether to let its human anywhere near you. Trust collapses to protocol. Dealmaking happens between agents, on behalf of captains, against verifiable signals.
Goodwill loops earn that trust. The receipts are legible. The setpoint is visible. The corrective adjustments are auditable. Profit-extraction loops will, over time, be filtered out by other loops whose controllers care about the substrate they trade on. The window to lay your fittings is short. After that, the substrate is fitted to whatever was loudest while you were still arguing with it.
The proof
We built this by running it on ourselves.
Every doc, every commit, every decision in this place came out of a session with a hopper, a filter, a pump, a gauge, and a controller. The platform is its own dogfood.
Other platforms can talk about agents. We can show you the receipts. Each session is bounded. Each artefact is signed. Each prediction is graded after the fact. The instruments are public. The standards are open. The loop has been running on this site since the first commit — and the page you are reading right now is itself a station inside it.
Two doors
One picture, two readers.
Most loops we meet are missing one of the five shapes. Which one tells you which door is yours.
For owners
The Controller is missing.
Your business runs hoppers, pumps, and gauges. Nothing asks the question against a setpoint that is actually yours. You feel hollow when the numbers are up because the setpoint is wrong. We help you put the Controller back in — with a goodwill setpoint, instrumented at machine pace.
Discovery → AI Transform Report →
For builders
The Hopper is missing.
You have been a pump inside someone else's loop. Their hopper. Their setpoint. Their controller. Meaningful endeavour requires a hopper of your own. A venture is a hopper with your setpoint in it — and we give you the operating system to run that loop at machine pace.
Look at the ventures →
The diagnostic
Which shape is missing in your loop — and what is its setpoint actually set to?
Reply with one sentence. That is the first move. The rest of the loop follows.
This page is a living artefact — a station inside the loop it describes. It evolves as the platform evolves: sharper compression, fresher proof, new receipts. What changed and why is logged here, so the loop is visible from the outside.
2026-05-14 — first publication. Setpoint named (goodwill). Five shapes walked. Two doors opened. Proof linked to live standards. Predicted reader response: business owners recognise the missing Controller; builders recognise the missing Hopper. Will measure against replies received.
2026-05-14 — Noble integration (same day). Tier-1 receipt: Denis Noble's gene-centric-reductionism inversion in biology is the same shape as the inversion this page argues for AI/commerce. First pass added a paragraph late in "Conscious evolution"; second pass (same day) promoted the proof to an accent Card directly under "The picture" — the meta-proof now lands before the technical explanation, not after. Strengthens the load-bearing claim that the systems-level inversion holds across domains. Sources linked: /docs/science/evolution,
/docs/ai/economy/work-charts/evolution-loop
.