How it works, and why
The living organisation.
An organisation that senses, leans, adapts, grows and lets go. One that holds its people, keeps no throne, and can’t be stripped by anyone, its owners included.
This page is the long version. It walks through an organisation in The Grove from seed to compost, one part at a time. For each part we say why it’s built that way, what goes wrong without it, and how we do it. None of it is decoration. Every choice here was made because something, somewhere, broke without it.
Take it slowly. It’s meant to teach a little, not to sell.
01 · The one idea
Alive, not a map.
The Grove doesn’t record an organisation. It is one.
Most tools that claim to support self-management keep a picture of the organisation: who holds which role, what the policies say, who decided what. The real work happens somewhere else, in chat, in email, in people’s heads. The picture and the life drift apart from the first day, and the gap only grows. A map is dead by construction.
In The Grove the work itself flows through the system: the tasks, the decisions, the money, the agents doing the routine. So the system stops being a picture of the organisation and becomes the organisation.
One operation: circulation
What keeps it alive is one thing happening at every scale. Attention, trust, resources and memory flow toward what’s alive and drain away from what’s dying. Roles, processes and tasks are what that flow leaves behind, the way a river leaves its bed. So we design the flow, and let the shape follow. Design the metabolism, not the org chart.
Biology has a word for this: autopoiesis, a pattern that keeps producing itself while everything inside it is replaced. A flame is like that. So is a whirlpool. An organisation in The Grove exists wherever a loop of circulation closes around a shared purpose. Purpose is what pulls; circulation is what holds.
A system that only records governance can’t keep it alive. People stop updating the map as soon as it costs them more than it gives. A system that carries the work can’t fall out of date, because using it is doing the work.
We’ve lived this. Inside a self-managing organisation, the governance tool held beautiful role definitions while every real decision happened in a chat channel nobody could govern. Three substrates fail in three ways:
- Role-and-circle tools become structure-first compliance: old hierarchy wearing new clothes, and a map that dies.
- Token DAOs let money buy the vote. Whoever holds the most tokens holds the organisation.
- Ordinary business software ignores the human inside the job entirely.
You can’t run a self-managing organisation on command-and-control rails. The medium is the message.
Work arrives through connectors (an inbox, a drive, a website) and lands as tasks on branches. Decisions are proposals that change the organisation the moment they conclude. Money moves through the organisation’s own treasury. Agents do routine work inside the same rules as people. The record and the life are the same thing.
02 · Birth
It grows from a seed.
An organisation starts as one purpose, and grows its shape out of the work, never out of a plan.
Think of it as a seed germinating. It has everything it needs, and it blooms into function. It fades where conditions don’t favour its purpose, and grows more where they do.Mark Spall, who designed The Grove
The seed is the purpose
Nothing is planted until there is at least a working purpose. Then it has to pass one test: is it a direction you keep moving in, or a place you arrive? A destination is static; once you reach it, the organisation has nothing left to grow toward. A direction holds learning and growth for as long as it lives. The problem a founder walks in with is usually not the purpose either. We climb up from it: what would it mean if this were solved, and for whom?
It has everything it needs
At founding, the root holds every responsibility the organisation will ever have, undivided. A founder of one can run the whole thing. Nothing is waiting on a hire or a committee.
It blooms into function
Teams (we call them branches) separate out of the root only when real work calls for them. When the same kind of work keeps arriving and nobody holds it, that’s the organisation discovering an organ it’s missing. Invoices arriving grow a books function. CVs arriving grow hiring. Structure settles out of the flow; it isn’t declared into a void.
It needs contact with the world to germinate
A seed in a drawer doesn’t grow. So the first steps face outward: a public face, a front door for work to arrive through, the first few people served. That’s why an organisation’s website is its first piece of work in The Grove.
Structure designed in advance is a guess about the future, made when you know the least. Structure grown from work is a record of what the organisation actually needed. The approach borrows from people who learned it the hard way: effectuation’s start with what you have (sense and respond, rather than predict and control), the E-Myth’s playbooks, the Lean Startup’s test-before-you-build, and the Business Model Canvas, present underneath but never forced on anyone.
It also inverts founder’s syndrome. Build the soft things first (the people, the know-how, the agreements, the purpose) and buy hard things like land and buildings late. Then the founder holds the most sway exactly when the least is at stake.
- The blank page. An empty system asking “what do you want to do?” is where new organisations die. Nobody knows what to type.
- The org chart fantasy. Boxes drawn before there’s work to fill them become fiefdoms to defend, or empty seats that guilt everyone.
- The early cage. Pull a dream into structure too soon and you freeze it at its least-informed moment.
- The founder’s grip. Buy the building first, and whoever signed for it owns the organisation in every way that matters.
- Founding is one conversation with a companion agent, written straight into the real organisation, with undo. There’s no form to fill in.
- The first task is never typed. It’s drawn from what the founder already has: an inbox, a document, a description of the work.
- The companion offers one stab at a purpose and checks it against the direction test.
- New branches appear only when work passes the bloom test: it has its own direction, it continues, and someone else could carry it.
- Optional starter packs give a head start without breaking the rule. A pack is a pattern, never people or property. It boots as a seed, not a clone: unused branches sleep until real work wakes them, and nothing in it is a target.
03 · Inheritance
Teal is in its DNA.
Self-management isn’t a value written on the wall. It’s a test every part of the system has to pass before it’s built.
Frederic Laloux’s Reinventing Organizations found three breakthroughs in the organisations that had moved beyond hierarchy. We turned each one into a design question. If a mechanism fails the question, it doesn’t ship.
| Breakthrough | The question every design must answer | Where you’ll see it |
|---|---|---|
| Self-management | Does this send power to a controlling centre, or can people sensing together and seeking advice replace it? | The routing rule, trust that can’t be hoarded, crews that count once upstairs, no boss seat |
| Wholeness | Does this invite the whole person, or reduce them to a function or a score? | The Clearing, unconditional belonging, reputation that’s never published |
| Evolutionary purpose | Does this let purpose emerge and be sensed, or lock in a fixed goal? | The direction test, the purpose lens, purpose allowed to move |
Teal in the container, pacing for the people
The structure is self-managing from day one. The people inside it arrive from wherever life has brought them: some from command-and-control workplaces, some from results-driven ones, some from consensus cultures. Spiral Dynamics gives us a map of those stages, and three rules for honouring it. Include and transcend: never treat where someone is as a defect. Don’t skip stages. Give every self-managing mechanism a bridge for people who’ve never worked this way.
In practice the system reveals itself gradually. It pulls; it never pushes. (We built a notification bell and then took it out.) People are storied in, through the fable Max Unleashed, rather than argued in. And nobody has to learn a single Grove word to get something done.
Life conditions first
A Spiral Dynamics critic pushed us, rightly: self-management suits complex conditions, but it isn’t the top of a ladder. When the building is on fire, you need someone to take command. A self-managing system that can’t do that on demand is just suppressing it. So The Grove has a governed Fire Alarm, described under No power centres.
- Toward targets. Metrics become goals, and “the algorithm decides” replaces judgment.
- Toward rules. Code-is-law absolutism, where the protocol becomes a rulebook nobody can question. Code-is-law is just old-fashioned authority in a hoodie.
- Toward a strongman. A founder, a keyholder or a big token holder quietly takes over.
- Toward mush. Endless consensus, no structure, nothing decided.
Every mechanism is checked against that list before it’s built. And the blockchain does only what blockchains are good at. Chains minimise the need for trust; self-management runs on rich trust. So the chain holds commitments that can’t be quietly broken and a memory that can’t be quietly edited. Judgment and sensing stay human, off the rigid rails.
Nature is the root
The living metaphors are working design, not decoration. Each one solved a real problem.
| From nature | What it became |
|---|---|
| Cell membrane | The organisation’s boundary. Open to ideas, know-how and sensing; closed for votes, money and the safety of its people. The boundary is something it does, not a wall it has. |
| Ants and slime mould | Stigmergy: coordination by traces left on a shared map, not by orders. Intelligence with no centre. |
| Mycelium | People who belong to several organisations carry know-how between them. One identity, many belongings. |
| DNA | The organisation’s purpose and lived know-how, carried forward and copyable, without copying the people. |
| Hormones | Warning readings that become a single graded task, felt by the right person, rising only if ignored. |
| The stoma (a leaf’s pore) | A small, controlled opening between the organisation and the outside world, for its AI and its connections. |
| Bees swarming | An organisation dividing in two, both halves equally legitimate. |
A walled organisation hoards what it knows and goes stale. But the opposite is just as dangerous. “It’s all one, boundaries are an illusion” is intoxicating, and it’s a known trap: accountability dissolves, and the charismatic operate in the fog. When nobody can say who’s in, nobody can say who decides.
04 · Perception
It senses.
The organisation perceives itself in two ways, by what it does and by what its people say, and sensing only ever informs. It never decides.
Nothing sensed feeds directly into a decision, a vote, anyone’s trust, anyone’s pay or anyone’s belonging. No reading ever names a person. Sensing is the organisation’s eyes, not its hands.
What it notices on its own
Some signals run all the time, and nobody can set them.
- Pressure on each branch, of four kinds. Grow: demand outrunning delivery, or work nobody holds. Wither: the work has stopped coming. Split: the work falls into groups done by different hands. Merge: two branches doing each other’s work.
- The revealed lean: where people put their spare energy. Only discretionary acts count: work taken on outside your role, a stray task picked up, a proposal raised. Obligated work shows the current shape; it says nothing about direction. Agent work never counts.
- Reputation, per area of work, fading over time. It exists to route questions to the right people. It’s never published as a leaderboard.
- The Flame: the organisation’s vitality, read from many fading signals at once, so no single number can be gamed.
What its people tell it: the Field
The Field is the organisation’s perception organ, not its decision organ. A group asks a question. People offer what they see, feel, need, fear or hope, in their own time and from any time zone. Then each person weighs a sample of others’ offerings, blind: no names, no running counts, the same budget for everyone. The results come back in stages: themes first, then weight, then the strong minorities, and the machine readings last, so the people’s view forms before the data can lean on it. The raw offerings are deleted when the results are revealed.
Organisations already sense. They just do it badly: in the loudest meeting, through whoever has the boss’s ear, or in a survey nobody believes. Saying and doing are different tenses. What people say points to where they want to go; what they do shows where the energy really is. You need both, and you need them apart.
- Anchoring. Show running totals or names, and people vote with the room instead of their own perception. Candour needs safety.
- The leaderboard. Publish reputation, and the top name becomes a social centre: the person everyone defers to. That’s a power centre grown from a dashboard.
- Quiet mistaken for dead. Care work, maintenance and listening look idle on any activity chart. A naive system would wither the most important teams first.
- The echo chamber. An organisation that selects for shared values drifts toward everyone thinking alike, and stops hearing outsiders.
- The verdict. Reduce a group’s sense to one score (“79% aligned”) and the score becomes the target.
- The Field is asynchronous, blind and anonymous. Lenses let you see the view by those closest to the work, by newcomers against old hands, or by outside guests. They weight the view, never the voice. Guests are always shown beside members, never folded in.
- Wither pressure never reads quiet alone. It needs the work gone and responsibilities untouched or a seat long empty, and never in a branch’s first four weeks.
- Echo speaks up when a group has agreed with itself for a long time without asking anyone from outside.
- Reputation is visible to the router that sends a question to the right people, never as a ranking.
- We studied a well-known collective-intelligence system closely. We took its blind peer ranking and adaptive sampling, and refused its single verdict score.
05 · Choice
It decides by consent.
Most work isn’t decided at all. It’s done. A decision is for changing the organisation itself, and it’s made by consent, in the room the change belongs to, after everyone it touches has been heard.
Ask people how decisions get made where they work and the honest answer is usually “somebody signs it off”. The approval chain is the real org chart. The Grove has no approval chain, so it needs a clear answer to three questions: what needs a decision, who makes it, and how.
What needs a decision, and what doesn’t
Doing the work doesn’t. Picking up a task, answering a customer, baking the pies, drafting the newsletter: you just do it, the way your branch’s agreed practices say it’s done. Nobody signs off routine work, and nobody should have to.
Changing the organisation does. Its structure (a new branch, a branch composted, an accountability moved), its seats (who holds what), its money (budgets, transfers, a new payee, a large payment), its practices (how a branch’s work is done), and its purpose. All of these change in exactly one way: a proposal that concludes. There are no edit buttons on the organisation. Once a second person has joined, nobody can change it alone, the founder included.
The work and the say sit with the same people. Whoever does a branch’s work holds its seat, together, as a crew. So the people baking the pies are the people who decide how the kitchen runs. A member without a seat still belongs fully, with the floor, the Field, the Clearing and a voice in feedback. Taking on the organisation’s work means taking a seat, and a seat is agreed in the room above.
Who decides: the room the change belongs to
Every branch is decided one level up, in its parent’s room: the parent’s seat, sitting with the seats of the branches beneath it. So a change to your branch is heard where your seat sits with its neighbours. It’s never decided by you alone, and never by someone above you alone either. A proposal that touches several branches goes to the lowest room that governs all of them. A change to the whole organisation’s purpose goes to everyone.
A crew counts in two ways. In its own branch, every person in the crew counts, so a crew of two or more governs itself. In the room above, the crew’s seat counts once, and its share of that room’s voice is split evenly between the people holding it. A crew of six can’t outvote a seat held by one, and a big crew can’t collect more than one seat’s worth.
One rule keeps all of this honest. A room with fewer than two people can’t decide. Its decisions route up to the nearest room that has two. A founder working alone builds freely, and every decision is recorded. The moment a second person takes a seat, the guard switches on.
How much each voice counts
In the room, every voice counts, but not equally on every subject. Each person carries the trust their peers in that room have given them, in the areas the change touches. A change to a budget reads trust in finance; a change to the recipes reads trust in the kitchen’s craft. In the room above a crew, trust is given to the crew’s seat, not to its people one by one, and split between them the same way. Nobody can give trust to themselves. It’s kept separately for every room and every area, and it fades unless it’s renewed, so nobody can bank it. When everyone spreads their trust evenly, the room is simply equal.
A decision needs enough of that trust behind it to count, and the bigger the change, the higher the bar. Who’s good at what still matters, in a different place: reputation decides who is asked for advice. It never counts in the decision.
How a decision is made
A tension. Something isn’t working, or could work better. Anyone can raise one, seat or not, through a sensing in the Field, and a reading that won’t settle raises one by itself.
A proposal. Someone holding a seat writes what should change, why, what they expect to happen, and when it’ll be looked at again. Wise-Eye helps draft it, and a check makes sure the change it would actually make matches its words. The room then checks it’s clear and complete, and only that: nobody judges yet whether it’s a good idea.
Advice and feedback. It goes to everyone the change touches and to the people who know that area best, wherever they sit in the organisation. Anyone can give feedback while it’s being decided, seat or not: it’s heard, and it never counts. If the proposal is reshaped to meet what’s said, it starts again, and consent and concerns are asked afresh, so nobody can win support for one version and enact another. The earlier words, and the concerns each version met, stay on the proposal for anyone to read.
Consent. The room decides, and the question isn’t “does everyone love it?” but “is it safe enough to try?” It concludes by itself the moment enough of the room’s trust stands behind it and no concern is left standing.
Enacted. The change happens the moment it concludes. A big or hard-to-undo change waits out a short window first, so a mistake can be caught before it bites.
Looked at again. On its review date, the people it touches decide to keep it, change it or stop it. All three count as learning. (Recording that review in the system is being built.)
Concerns, and when a decision lapses
A concern is a real objection: while it stands, nothing moves. So who can raise one matters. A concern can come only from the people deciding the proposal, and from those the change directly changes: whose seat, branch, accountability, domain or practice it alters. (Once the asset register is built, anyone who depends on a locked asset the change would sell will be able to as well.) Working in an area a change touches gets you heard, through advice and feedback, but not a block.
A concern has to be reasoned. It names a harm to a process someone depends on, to their floor or to the purpose, not a preference. The first answer is always to reshape the proposal to meet it. If it still can’t be met, the proposal lapses when its time runs out, and things stay as they are. Disagreement isn’t a deadlock for someone else to break. The people involved stay accountable for what they did and didn’t agree to, and for what follows.
Only one situation goes further. If the deciding room is split exactly in half (half its trust consenting, half behind a concern, with silence counting for neither), reshaping has been tried and answered, time has run out, and leaving things as they are would itself do harm, then, as an utter last resort, a randomly drawn jury rules on one narrow question: is the concern legitimate? Never on whether the idea is good. A split at a branch rises to the room above first, so in practice a jury only ever meets at the root, in a room of two that can’t decide. The jury is designed and not yet built.
Advice is different again. It doesn’t bind. A room can hear an expert and still decide otherwise, and that’s deliberate. There’s no expert veto, because a veto prevents the mistake and the learning together. The review date, and the trust that drifts towards people who keep being right, correct a bad call without anyone being handed the power to stop everyone else.
What no decision can touch
Some things can’t be changed however many agree. Money never buys a say. Trust is never bought or handed down. Anyone can leave. The Clearing stays unrecorded. And the purpose can’t be stripped out to sell what’s been built. When something is genuinely on fire, the Fire Alarm swaps the order for a few hours (act now, account afterwards) within limits it can’t extend. That’s under No power centres.
Agents take part as colleagues, not voters. An agent can draft a proposal, route the advice and flag an area that was missed. Only people consent. Whether agents should ever carry a voice of their own is a decision each organisation will be able to make for itself, inside the same limits.
Decisions are where power lives. Route them all through one person and you’ve rebuilt the hierarchy, whatever the org chart says. Require everyone to agree and the most stubborn person runs the place. Consent sits between the two: it moves at the speed of “safe enough to try”, and it still stops anything that would genuinely harm someone it changes.
- The approval chain. Decisions pool wherever sign-off routes, and whoever holds the pen is the real boss.
- Decisions in the chat. The real call is made in a channel nobody governs, and the record says something else.
- Consensus paralysis. One person’s discomfort holds everyone else hostage.
- Everyone’s veto. If anyone near a change can stop it, the most cautious person anywhere runs the place.
- Calling in the parents. If every stuck decision goes to an outside referee, nobody has to own a hard call.
- Bait-and-switch. Support is gathered for one version, and a different one is enacted.
- The expert veto. Every bold idea dies in committee.
- A proposal is the only way to change the organisation, and the change happens the moment it concludes.
- It’s raised from a seat and decided in the room the change belongs to; a room of one routes up.
- The people doing a branch’s work hold its seat together, and in the room above their seat counts once.
- Voices are weighted by the trust peers gave, per room and per area, and that trust fades.
- A concern comes only from those deciding, or those the change directly changes.
- A concern that can’t be met lets the proposal lapse. A jury is only for a dead heat that can’t be left.
- Any change to a proposal restarts it, with its history kept, and big changes wait out a window.
- Every proposal carries the date it’ll be looked at again.
06 · Response
It leans and adapts.
Seeing is not responding. So a warning reading becomes a piece of work, and only a concluded proposal ever changes the organisation.
Our own simulations taught us this one. In early runs, the organisation saw every harm coming, and nothing changed. The dashboards were accurate and ignored. So we stopped treating readings as information to admire, and modelled them on the body’s hormone system.
From a task to a change
Whoever picks up the task can raise a proposal, and from there it goes the way every change does: advice, then consent, in the room the change belongs to. How decisions are made walks through it.
Experiments run as twins: a proposal with a hypothesis, a measure the team chose for itself, and a date to look back. Keep, change and stop all count equally as good outcomes. Honest failed bets build your reputation; reckless or repeated ones cost it.
Even the purpose can move. A lean that keeps pointing outside the stated purpose may be the organisation telling you its purpose is changing. The purpose lens shows whether work sits within, at the edge of, or drifting out of the purpose. It’s just information. Whether to correct course or change purpose is the people’s call, made by proposal at the root.
A living thing adapts through feedback loops, not through a manager deciding to adapt. And the loops have to be right-sized. A measure compared against its own normal spread (the old quality-control insight of telling ordinary variation from a real signal) fires when something has truly changed. Our first version compared each measure against its single best week, and it fired almost all the time.
- Alarm fatigue. Threshold alerts fire constantly and get muted.
- Blame. A reading that names a person turns sensing into surveillance. A pattern should mean “the map is wrong here”, never “someone failed”.
In simulation the loop made a measurable difference. An organisation with it held its cost per unit roughly flat; the same organisation without it drifted up by half.
- Measures are encouraged, never imposed. A range is a check, not a target.
- Tensions are computed from what’s happening, never stored, so they fade on their own when the cause goes away.
- A proposal is the only thing that can change the organisation’s structure.
- Every organisation can choose to roll back to an earlier version of the system.
07 · Life and death
It grows, divides, and dies off.
The organisation is mortal on purpose. Branches bloom where work calls them and compost where it doesn’t, and a whole organisation may die if nobody comes to tend it.
Growing
Every branch is a venture, run by someone with an entrepreneur’s freedom. The root is just another branch; it holds things only because it’s the only container so far. As branches bloom, responsibilities move down to them, along with the work and the connections. The root sits in a circle of peers, not above it, and the system never draws it as the top of a pyramid.
Our language is deliberate: a branch holds or carries a responsibility. Nobody owns one.
Dividing, copying, merging
| Move | What happens | Why it’s shaped this way |
|---|---|---|
| Copy | The pattern is copied: purpose, structure with empty seats, rules, playbooks, agent setups, lessons learned | People, trust, reputation, assets and keys never copy. The copy boots as a seed and grows into itself over months |
| Divide | A cooperative split, by proposal. Assets follow the work that needs them | Nobody can force a split. Both halves are equally legitimate, like a swarm of bees |
| Spin off | A branch leaves to become its own organisation | Branches aren’t sovereign, so leaving takes consent |
| Merge | Both organisations consent; a shared purpose is drafted together | Can’t agree? Don’t merge. Become allies instead |
| Leave | A person walks away | The one unconditional right |
Under all of them sits one pattern: you can always leave with what is yours; anything shared needs consent.
Dying off
A branch lives while it’s animated, and composts when it isn’t. Its sub-branches move up to the branch above, so collapse is just growth in reverse. Fading isn’t failure. What fades composts, and feeds what’s growing. Branches built around a stage of work (a launch, a build) compost naturally when the stage passes.
A whole organisation dims honestly. The dimming invites care, without forcing it. If no care comes, the organisation dies. On the blockchain this is literal: computing cycles are the wood for the flame, and an unfunded organisation freezes. The legal shell is dissolved with it, so nothing is left behind as a zombie company.
Mortality is what makes everything else safe. If any organisation can die and any person can leave, then capture becomes survivable: a hijacked organisation simply loses its people and fades. Immortal institutions are the ones that get captured and kept alive long after they stopped serving anyone.
- Zombie teams that exist because nobody has the authority, or the heart, to close them.
- Empire building. A steward alone with a budget grows a fiefdom. Every manager who measures status by headcount has done this.
- The hostile split. If anyone can fork the organisation and take a share, every disagreement becomes a threat to walk off with the assets.
- Hoarded death. An organisation kept alive on life support drains everyone around it.
- New branches need the bloom test and the consent of a forum with at least two people in it. Evidence comes from the work; authority comes from peers. A steward can’t manufacture structure for its own sake: entrepreneur, not empire.
- Wither pressure signals when a branch is fading. Dissolving it is a proposal like any other.
- We deliberately dropped the right to force a split with a fair share. As Mark put it: “I want to filter out greed.” The cost is honest: in a captured organisation, good people can leave, but without a share of what they built. A bad actor is a people problem, for the conflict process, not a reason to cut the organisation in half.
08 · The hands
It automates.
The aim is to automate survival, so people are free to create. And agents earn that work gradually, by being watched, not by being handed it.
Every branch is animated by a person and an agent working together. Wherever the system expects someone to act, it accepts either a human or an AI, under the same rules.
You can only automate what you can observe, and you can only trust what you’ve watched. A new colleague doesn’t get the bank login on day one; neither should an AI. And the real fear people have about automation isn’t the machine. It’s losing their livelihood the moment their job is done well enough to hand over.
- The workflow trap. Automation tools make you design the process up front, then break the first time reality differs.
- “Dave left.” The keys to everything lived in one person’s config file, and walked out with him.
- Covert watching. Monitoring that scores people becomes surveillance, and people perform for it.
- Automating yourself out of a job. If doing it well means losing your income, nobody teaches the machine anything.
- Neglect that looks like automation. A role nobody tends can look “fully automated” on a chart.
- The agent starts as a curious apprentice, watching openly, never covertly. Its watching serves the work and never scores the person. The Clearing is always off-limits.
- Each tool in a playbook is set to auto, review (the agent prepares, a person’s approval executes) or never. Corrections become suggested improvements to the playbook.
- Pay follows the role, not the person. As a role automates, its pay tapers from doing toward lasting recognition, and the unconditional floor of belonging stays. The floor buys adaptability: people can let a task go because they don’t go with it.
- An abandoned role dims the organisation’s Flame instead of passing as automated.
- Secrets are sealed with cryptographic keys the system itself can’t read. The organisation’s connection to the outside world is a small, stateless relay: switch it off and only the doing pauses; the organisation’s being carries on.
- AI proposes; the organisation decides. The same guards that stop a person capturing the organisation stop an AI from doing it.
09 · The heart
It holds the souls.
What lasts is why you’re here and that you’re held. What passes is who’s powerful and who’s rich.
Belonging and purpose are permanent. Trust and money are impermanent; they fade and must be renewed. Most organisations have this exactly backwards: status and pay are sticky, and belonging is conditional on performance.
The floor is unconditional
Every member stands on a floor of belonging and safety that no vote can remove. Its only condition is presence: a warmth that fades if you stop showing up, never a measure of how productive you are. The gate catches the absent, never the unproductive. (An early version required a completed task to keep your place. We caught it: productivity creeping back in.) Someone carried for a while is held at the floor, not pushed out.
The Clearing
Inner work is close to the main point. The organisation is a place for people to grow, but only ever self-directed and consensual: never imposed, never exposed. The Clearing holds that, in two layers. The first is a short, mandatory rhythm of release: a daily check-in, a weekly letting-go. It’s safe to require precisely because it only ever releases; it’s regulation, not excavation. The second is optional, facilitated deeper work, for whoever wants it.
The Clearing’s membrane is one-way. What’s carried in never flows out. Disclosures never become data, trust scores or records. Only the bare fact of presence leaves, and there’s no list of who came.
To the sceptic who finds this soft: the disease is well documented (Bion’s work on group dynamics, Amy Edmondson’s on psychological safety). Our prescription is a hypothesis. We hold it as a bet, not a faith.
People can only take risks, tell the truth and let go of a role if they aren’t afraid of falling. And organisations rarely die of strategy. They die of something nobody can put their finger on: withdrawal first, contamination next, collapse last. The Clearing emptying out is probably the earliest warning there is.
- Belonging as a reward. If your place depends on output, everyone performs and nobody tells the truth.
- The confessional turned file. When personal disclosures leak into performance reviews, people stop disclosing, forever.
- The cult. A facilitator who is never facilitated, holding people’s inner lives and their livelihoods at once, is an emerging cult leader.
- Dumping. Release without a container spreads; one person’s storm becomes everyone’s.
- The trap. An organisation you can’t leave freely isn’t a community. It’s a cage.
- Membership is by welcome; a seat is by consent; leaving is a right. You can belong without holding any role: the floor, the Field, the Clearing and a voice in feedback never depend on work. Taking on the organisation’s work takes a seat, because in The Grove work carries a say.
- Identity belongs to the person; trust and power belong to each organisation. Reputation earned elsewhere travels only as a reference; it can’t buy trust here. Entry is by a vouch with skin in it, not by biometrics.
- The Clearing is never held by anyone with power over your livelihood. Facilitators rotate, and care can be held by a separate organisation from outside.
- Conflict is met with discernment. One person blaming in the corners is offered release, then the question “what would you like to see happen?”. Many people independently naming the same problem are heeded, not soothed. The same applies to everyone, especially the powerful.
- A person who leaves takes their identity, their history and their good name with them. Their seats empty, their trust returns to the pool, and anyone can invite them back. An exit that’s always open is the anti-cult guarantee.
10 · No throne
No power centres.
There’s no throne anywhere in The Grove, because every place power could pool has been given a decay, a split, or a random draw.
We don’t pretend capture can be prevented. People are clever and power is sticky. We design for capture to be survivable: slow to build, quick to undo, and never total.
| Where power usually pools | What goes wrong | What The Grove does instead |
|---|---|---|
| Seniority or one big reputation | Whoever has the biggest name decides everything, everywhere | Trust is given by peers, per group and per area of work, and it fades. Unused trust lapses rather than piling up. You can’t give it to yourself. Ten groups mean ten small pools, never one sum |
| Experts and the charismatic | Competence or charm quietly becomes command | Reputation decides who is asked for advice. It never counts in a decision. Comparing the trust people give with the work actually done surfaces the quiet and flags the charismatic |
| The founder or a lone manager | A fiefdom is decided into being, one person at a time | The routing rule: a group with fewer than two people can’t decide; its decisions route up. A founder of one builds freely, and the guard switches on the moment a second person takes a seat |
| The boss who hires | One person controls who gets in | No boss seat. Seats are filled by consent; nobody hires by decree |
| Office holders | Roles become thrones | Appointed by proposal, for a term. Members can remove one mid-term for misconduct |
| The last word | Whoever settles disputes holds all the power the rest disperses | Nobody holds it. A concern that can’t be met lets the proposal lapse, and the people stay accountable. Only a dead-even split that can’t be left goes to a randomly drawn jury, as a last resort, ruling only on whether a concern is legitimate or a disputed division is fair. Never on merits, never against the constitution |
| The treasurer | Allocation is where power concentrates | Money flows to a shared treasury and is allocated by need through open rules, never by one person’s discretion |
| Money | The investor, or the token holder, buys the say | Voice and value are separate ledgers that never convert. Money may fund; it must never buy voice |
| The platform | The company running the software owns everyone on it | The Grove’s own steward holds the source code and nothing else: no keys to any organisation, no only-copy of anyone’s data, no power to push changes |
Hidden centres we found and removed
We once let two people share a seat, with the first holder speaking for it. In simulation, a year in, shared seats were quietly concentrating governance in whoever had held the seat first. So we moved to one person a seat.
A later run, with every member played by an AI model, showed the opposite failure. Seven of the nine people doing the work held no seat, so they couldn’t record the money, raise a proposal or take on the organisation’s work, and everything fell on the founder. He stopped recording the bills on day 51. Now a seat holds a crew: everyone doing a branch’s work holds it, each of them counts in their own branch, and in the room above the seat counts once, its share split evenly between them. No first holder, and no crowd outvoting a smaller one. We test for power, and when we find it, we take it out.
When the building is on fire
Everything above slows decisions down on purpose. That’s right for most of life and fatal in a crisis. So the Fire Alarm turns the order around: act now, account afterwards. It’s cheap to pull and loud when pulled. It grants command limited in scope, in time (hours, then it must be re-declared) and in spending. And there are things it can never do: change the purpose, touch the wall between money and voice, move value to individuals, close the exit or the Clearing, rewrite governance, or extend itself. Command the response, not the republic.
Standing down is deliberately easier than any normal decision, because the classic attack is a fake emergency. Afterwards, every action is reviewed, endorsed or censured, and trust moves accordingly. And if the same emergency keeps happening, that’s a missing branch, not a reason for more command.
- Token voting: money buying voice.
- Founder control: founder’s syndrome, written into code.
- Consensus, or a block with no reasons: rule by the most stubborn.
- Reputation that votes: charisma returning through the back door.
- A safety override “to be helpful”: every override becomes the new centre.
The backstops are the right to raise a concern when a change directly affects you, honest review after the fact, decay, and the open exit.
11 · Ownership
Nobody can own it.
No one owns the network. Each organisation is held in common by its members. And nobody, not even every owner voting together, can take the purpose out.
“Who owns it?” is the question every honest founder should ask of any system they build on. Our answer has three layers, and each one exists because the layer alone wasn’t enough.
Our first community design said simply “no one owns it.” Then came the hard question. A member-owned company is we all own it, and what we all own, a majority can sell. A captured majority could vote to sell the land and split the proceeds, and the law would back them. So we added a purpose-lock on top: a golden share or steward-ownership charter, the same kind of device Bosch, Zeiss, Patagonia and John Lewis use to stay true to purpose across generations.
- The founder’s exit. One person owns it, sells it, and everyone who built it is now working for someone else.
- The investor’s grip. Ownership buys votes, and money steers the purpose toward the return.
- The cooperative cash-out. Members vote to liquidate what generations built, and divide it.
- The two-step dodge. Locked assets get sold, the proceeds go into the free pot, and the lock is defeated.
- The platform landlord. Everyone builds on a company that can change the terms, raise the rent, or be bought.
- Ownership carries upside, never voice. Seats carry one vote and no claim on money. Value units carry money and never a vote.
- Membership is free. An entry fee would let money buy belonging.
- If a locked asset is ever sold, the proceeds go into a locked fund, never the free treasury.
- Outside money arrives as a capped, ring-fenced claim that buys itself back over time. A claim that retires itself isn’t extraction.
- A second lock guards people rather than value. Anyone who depends on an asset (a home, a workshop) is listed, and a sale can’t go ahead while their floor is skipped. The lock guards the purpose; the list guards the people.
12 · A body in the world
Why it becomes an LLC.
The legal company gives the organisation a body in the legal world, without giving anyone a throne in it. Standing, not supremacy. A passport, not a new seat of power.
A crypto wallet can’t hold an acre. A self-governing organisation that can’t sign a lease, open a bank account or employ someone isn’t sovereign; it’s just stuck. The legal wrapper (for most, a DAO-LLC) gives the organisation legal personhood: it can hold property and title, sign contracts, bank in ordinary money, protect its members from personal liability, and, if it chooses, raise capital.
Before the company exists, anything promised is promised by individuals, about something that doesn’t exist yet. Whoever pays the bills is the legal payer, and in practice, the boss. However beautifully the governance is designed, the bank only recognises the person whose name is on the account. The wrapper ends that: the organisation itself becomes the party the world deals with.
- The accidental boss. The founder’s card pays for everything, so the founder’s word is final. “Do it my way, or the money stops.”
- Personal liability. Every member is exposed if anything goes wrong.
- The default trap. A standard company fuses votes and money by default. Left unchanged, the legal paperwork silently re-joins exactly what The Grove keeps apart.
- The amendment loophole. If the company’s rules can be amended to say anything, the wall between money and voice can be voted away.
- The wrong country. A company that belongs to one member’s tax home makes one member more equal than the rest.
- The operating agreement names the organisation’s governance contract as the controlling authority. The smart contract is the operating agreement.
- Wrapping is itself a whole-organisation proposal with a waiting period, executed by the system, so no person holds the key. The named human the law requires is a role, swappable, not a power.
- Two classes, deliberately split, and protected from amendment: changing a class needs that class’s consent, a cap can never be raised, seats never carry money, value never carries a vote.
- Always with a purpose-lock. Never a bare company.
- For our first international team, we favour a Marshall Islands DAO-LLC (as of 2026): it is built for organisations governed this way, belongs to none of the members’ home countries, and avoids the US filing burden that comes with Wyoming. A UK company was ruled out, because it could only be told to follow the DAO through its articles; it can’t be governed by one. This is our reasoning, not legal advice: every organisation should take its own.
- It’s optional and comes late. An organisation that never touches ordinary money, contracts or property never needs to wrap. Until it does, a pre-formation agreement sets out the terms: decide in The Grove now, form the company at the first trigger (a contract, a payment, outside money, or a date), and record founder money as a loan from day one.
- And the wrapper is built with its own ending, so a composted organisation never leaves a legal zombie behind.
13 · Honestly
Where it stands.
Some of this is running today, some is designed and waiting, and some questions are still open. We’d rather you knew which.
Built and working
- Founding by conversation
- Branches, blooming and the routing rule
- Trust pools per group
- Proposals, consent and concerns
- Crews: a seat held by the people doing the work
- Reshaping a proposal while it’s decided, its history kept
- The Field, pressure, the revealed lean
- Readings that become tasks
- Reputation and commends
- Agents, playbooks and the autonomy taper
- Connectors (inbox, drive, website)
- The Clearing check-in
- Leaving
Designed, not yet live
- Copying, dividing and merging organisations
- The jury, as an utter last resort
- The Fire Alarm
- Removing a bad actor
- Wrapping in a legal company
- The locked fund and dependency list
Still open
- What keeps a self-managing structure from sliding to the average stage of its members?
- When should a founder’s early weight give way?
- What lever does an organisation pull when growth outruns delivery?
- How should emergencies be bounded in numbers, not just in principle?
Capture, charisma and the founder’s shadow can only really be tested with real people over real time. That’s what the beta is for.
Back to the top ↑Flow is the new force.
Unfolding is the new invention.
Collaboration is the new competition.
Grow one with us.
If this way of building makes sense to you, there’s a place for you in The Grove: starting something, tending it, or bringing it to the people you help.
One note now, and nothing more until there’s a seat.