The AI Company Founding Thesis · Intelligent Internet

AI is becoming part of how every economy works. The people who depend on it should have a say in its direction and a route to own the institutions delivering it. The AI Company is a model for a public-benefit company, majority-owned by local people and institutions. Each local company is a Champion.

Part I · The Change and the Institution

Intelligence throughout economic life

AI is moving from answering questions to carrying out work. As its abilities grow, the opportunity extends through software, institutions and the physical world.

Three capabilities, one opportunity

Understand, arrange, carry out

Connecting these capabilities opens a much larger opportunity than any single application.

Personal agentunderstands a need
Organisationarranges the response
Machinecarries out the physical task

The ownership decision is being made now

2028 · The Last Economy forecasts the economic value of human cognitive labour turns negative: retaining human execution will cost more than the available machine substitute.

Figure 1

A personal agent can understand a need, an organisation can arrange the response, and a machine can carry out the physical task. The Last Economy forecasts the value of human cognitive labour turns negative in 2028.

Where it's heading

The ownership decision is being made now

As wages become a less reliable route to the value produced by intelligence, ownership matters more. Universal access helps people use better tools; an ownership interest lets them participate in the business those tools make possible.

Humanoid form gives intelligent machines a route into places built for people. The same customer relationship extends from digital services to physical work.

The takeawayPut improving intelligence to work. Give local people a share in what it creates.

Part I · The Change and the Institution

Own the institution that gets the work done

The last mile of intelligence connects AI to a real need, a working relationship and someone responsible for the result. The Champion organises that connection.

The last mile

Global suppliers, local responsibility

The Champion holds the customer relationship and accepts the obligation to deliver.

Figure 2

Personal agents serve people in everyday life; engineers and domain specialists work directly with organisations; intelligent machines extend those services into the physical world. Together they bring recurring demand for models, computing, software, expertise and equipment.

Local responsibility, global suppliers

Language, law, professional duties and existing systems shape how work is done. The programme establishes one Champion in each participating jurisdiction: a US state, or the appropriate country or region elsewhere. Customers choose their providers, and public institutions retain their authority. Each Champion takes the name of the place it serves: the AI Company of California, or the AI Company of the UK.

The takeawayThe Champion connects global suppliers to local customers and carries responsibility for delivery.

Part I · The Change and the Institution

An agent for everyone. Capability for institutions.

Universal access and the work of public institutions anchor the mission. The local team builds a commercial business serving needs throughout its economy.

The public mission

An agent for everyone. Capability for institutions.

Every citizen should have a competent personal agent to help them learn, plan, create and complete tasks.

0 of 96 citizens with an agent

  • Governmentadministration, policy evidence, citizen assistance
  • Courtssource-linked research, case preparation
  • Medicine and careevidence, records and operations
  • Educationlearning support, teacher tools, curricula

Founding capital establishes the universal service; income from paid services is intended to sustain it.

Figure 3

The basic service defines what the agent can do, its supported languages, included usage, availability and user support. Businesses use the same foundations for research, engineering, administration, customer service and production.

One foundation, many useful services

A person controls the purposes, permissions and information that let their agent act for them over time. The Champion integrates into existing organisations, works with local partners and competes for business on the quality of its service.

Work the company can support
InstitutionWork the company can support
GovernmentAdministrative workflows, policy evidence, public-service coordination and citizen assistance.
Courts and the judiciarySource-linked research, case preparation, document workflows and court administration.
Medicine and careEvidence, research, records and operations, with clinical use directed by the responsible professionals.
EducationLearning support, teacher tools, local curricula and school and university operations.

The takeawayPublic bodies, professionals and customers retain the decisions entrusted to them. The Champion coordinates the work and takes responsibility for what it contracts to deliver.

Part II · Demand, Ownership and Formation

Demand and knowledge make the company stronger

The Champion brings demand together where intelligence is used. It earns by delivering useful work and improves through what that work teaches it.

Demand brought together

Personal agents, organisations and machines generate recurring demand for AI, computing, software and support. Bringing these needs together gives the company a stronger basis for purchasing.

  • Personal agents
  • Public institutions
  • Businesses
  • Intelligent machines

Demand improves purchasing and use of capacity. The Champion brings organised demand to AI developers, computing providers, cloud platforms and equipment suppliers, and takes responsibility for delivery.

Figure 4

Personal agents, organisations and machines generate recurring demand for AI, computing, software and support. Brought together, that demand gives the company a stronger basis for purchasing, planning capacity and maintaining service.

The Champion answer

A business built on delivery

Revenue comes from deployment and integration fees, ongoing service contracts, paid usage beyond the universal service, and the operation and support of intelligent machines. Local people benefit through useful services, shared knowledge, reinvestment, dividends when declared and the shares they hold.

Customers pay for work completed to an agreed standard and for a service they can depend on.
Founding Thesis, p. 6

The takeawayDemand improves purchasing and use of capacity. Reviewed experience improves the next delivery.

Part II · Demand, Ownership and Formation

Assemble the institution. Own the value you help create.

A major company can begin by bringing existing technology, skilled people, capital and customer demand together around a clear economic purpose.

The formation principle

Assemble the institution

A major company can begin by bringing existing technology, skilled people, capital and customer demand together around a clear economic purpose.

1987 · Taiwan

A new institution put existing resources to work in a new business model.

Figure 5

In 1987, Taiwan's ITRI transferred factories, equipment, technology and 98 professionals into the newly established TSMC. The Champion applies the same formation principle to intelligence.

A $1 starting valuation

The company begins at a nominal $1 pre-money valuation, or local-currency equivalent. One nominal founder share precedes subscriptions and the disclosed founding interests. Founding participants enter as the business is being established and share in the value their contributions help create.

The takeawayThe founding bargain brings the resources together and gives the people building the company a stake in its success.

Part II · Demand, Ownership and Formation

A council that helps create the company

The founding council brings together people who can secure customers, recruit leadership, commit resources and establish confidence in the institution.

The proposed founding arrangement

Who holds the company at founding

Local subscriptions fund the company; the founding participants supply the people and relationships that make it useful.

  • Local subscribers, founding council and builderslocal people and institutions retain majority ownership
  • Childrenheld in trust for eligible local people aged 18 and under at founding
  • Intelligent Internetfounding equity for establishment work, with paid pro rata rights

The children's trust over time modelled

10.0% at founding + 0 years

10% of founding equity at founding, with a further 0.5% allocated annually for newborn cohorts. The children's trust extends ownership across generations.

Figure 6

Local people and institutions retain majority ownership. A trust holds 10% of founding equity for children, with a further 0.5% allocated annually for newborn cohorts. Intelligent Internet receives 10% for establishment work.

A clear founding job

The council identifies founding leadership, secures initial customer commitments and agrees the formation programme. The elected board and operating leadership carry the continuing responsibilities.

The proposed founding arrangement
ParticipantThe proposed founding arrangement
Local subscribersShares reflecting subscriptions; local people and institutions retain majority ownership.
ChildrenA trust holds 10% of founding equity collectively for eligible local people aged 18 and under at founding, with a further 0.5% allocated annually for newborn cohorts.
Founding councilA defined founding and governance role, with paid pro rata rights in eligible later investment rounds.
BuildersShares earned through work under the company's compensation arrangements.
Intelligent Internet10% founding equity for establishment work, with paid pro rata rights in eligible later rounds.

The takeawayUniversal access reaches people whether or not they invest. The children's trust extends ownership across generations. The company must remain answerable to the people it serves as well as to its shareholders.

Part II · Demand, Ownership and Formation

Turn interest in AI into a productive institution

Demand to own part of the AI economy finances a company that serves demand to use AI. Local ownership is established first; strategic resources equip the launch; public markets broaden participation.

Demand to own part of the AI economy finances a company that serves demand to use AI.

R1 · Local ownership$1 pre-money
Who
Local people and institutions, subscribing under disclosed founding terms.
Purpose
The first $10m funds recruitment and the initial team. The round establishes the ownership base and finances formation and launch preparation.
Key term
Local ownership is established first.

Launch. After both rounds, the company launches its universal agent and initial institutional services. Local knowledge, training and shared research programmes develop alongside paid work.

Procurement follows the services the Champion needs to deliver, with choice among suppliers.

Figure 7

Three rounds in sequence. R1 establishes local ownership under disclosed founding terms. R2 follows immediately, for 10% of the company after the investment, with new cash capped at the amount raised in R1. After launch, a public listing can broaden participation once listing requirements are met.

Launch · Begin useful services

After both rounds, the company launches its universal agent and initial institutional services. Once listing requirements are met, a public listing can broaden participation; local majority ownership is preserved as participation grows.

The takeawayA $100m R1 followed by a $100m R2 gives a $1bn post-money valuation. Local majority ownership is preserved as participation grows.

Run the numbers

The rounds, in figures you can move.

The calculator applies the document's own rule for R2. Move the inputs and watch the valuation respond. Modelled figures are worked examples, not results.

The founding rounds, sized

Two rounds, one rule.

modelled
$1bnpost-money valuation after R2 modelled
$900mimplied valuation before R2
$200mtotal cash raised, R1 + R2

The rule from the document: R2 is for 10% of the company after the investment, with new cash capped at the amount raised in R1. Paid council and II pro rata subscriptions count inside the cap.

The company begins at a $1 pre-money valuation. Local ownership is established first; strategic resources equip the launch; public markets broaden participation.

Part III · Intelligent Internet

The technology and support behind the founding

II supplies the shared technology, expertise and establishment support that give each Champion a stronger beginning.

The technology

Products, and two thin layers beneath them

What people use on top; coordination for work and for knowledge underneath. Select a layer to open it.

The platform connects

  • II technologythe products and the two coordination layers
  • Partner productsmodels, computing, equipment and specialist services
  • Existing softwarethe systems organisations already run
  • Open and proprietary systemsboth, by design

Shared software and methods improve through successive launches.

Figure 8

The platform connects II technology, partner products and existing software, including open and proprietary systems. Zenith coordinates work; Radiant coordinates knowledge; the products put both to use.

The Champion answer

Launch and continuing support

II provides software, models, datasets, deployment methods and training, supports recruitment and helps establish the local team. Each Champion raises its own capital and is responsible for customers, staff and delivery.

The takeawayEach Champion raises its own capital and is responsible for customers, staff and delivery. II receives 10% founding equity and paid pro rata investment rights. The shared software and methods improve through successive launches.

Leading results

The same model, fifth to first.

FrontierSWE V1 · 17 tasks

Same model, fifth to first

The base model does not change. The coordination layer around it does.

  • GPT-5.5 · default harness5th
  • GPT-5.5 · with Zenith1st

Reported by II for its published 17-task FrontierSWE V1 comparison. Same base model, GPT-5.5.

Figure 9

In II's published 17-task FrontierSWE V1 comparison, Zenith moved GPT-5.5 from fifth to first using the same base model.

Part IV · Operating Intelligence and Earning Trust

An AI Company operated with AI

The Champion should be its own first deployment. AI helps the team organise work, retain knowledge and make the institution understandable to the people who own and use it.

Its own first deployment

The company runs on its own agents

Six kinds of work, one human gate. The highlight cycles on its own, or click a step to walk it yourself.

1/6Organise projectswhat the company has promised

A clear view of the company

  • What it has promisedCommitments, tracked as they are made and met.
  • What resources are availablePeople, computing, equipment and funds, kept current.
  • Where progress has stalledSurfaced early, with the reason.
  • Which decisions need attentionRouted to the responsible owner, with what a proposal means and what follows from it.

Agents act within granted authority. Delegating work cannot enlarge it. Verified identity, defined permissions, spending limits and the ability to withdraw access set the boundaries; actions remain traceable.

Figure 10

Agents organise projects, build and test software, coordinate suppliers, prepare accounts and track commitments. They act within granted authority; consequential decisions have responsible owners.

An academy tied to real work

Engineers working directly with customers establish services alongside local specialists. The academy develops people who understand a domain, direct intelligent systems and exercise judgement when work takes an unexpected turn. Software can be copied quickly; capable operators must gain experience.

The takeawayUse intelligence to make the institution capable, understandable and responsive.

Part IV · Operating Intelligence and Earning Trust

Intelligence that understands its place

Effective intelligence depends on understanding the world in which it acts. Language, duties, relationships and local knowledge are part of that understanding.

Who decides, and who answers for it

Influence and investment do not confer public authority

Each role keeps the decisions entrusted to it. The highlight moves on its own; click a role to hold it.

A personsets the purposes and permissions of their agent.

Shared knowledge can concentrate power. People need ways to inspect the assumptions a system carries, challenge them and correct mistakes.

Figure 11

A person sets the purposes and permissions of their agent. Courts, clinicians, educators and public authorities retain the decisions entrusted to them. Shareholders govern through the founding rules and elected board.

The Champion answer

Trust earned through conduct

Competent service, clear responsibility and a fair response when something goes wrong earn trust. People need practical rights to retrieve and transfer records they are entitled to use; ownership, a say in decisions and the ability to switch providers give them control.

Influence and investment do not confer public authority.
Founding Thesis, p. 12

The takeawayOwnership, a say in decisions and the ability to switch providers give people control over their relationship with the institution.

The frameworks

Intelligent Economics, the paper behind the framework.

Download the paper (PDF)

Part V · Compute, Commons and Foundation Coin

Make more capacity productive

Combining immediate work, flexible tasks and commons programmes, shared research, models and resources, helps the Champion use suitable computing capacity more effectively.

The compute economy modelled

Make more capacity productive

0% of capacity in use

  • Live servicescapacity matched to required quality, availability, privacy and response time
  • Flexible commercial workapproved tasks completed within an agreed window
  • Commons programmesresearch, curation and model work on contributed or purchased computing

Some work must happen immediately; some can wait. The mix of work creates opportunities to fill gaps in demand. Illustrative shape, not a measurement.

Figure 12

Live services run on time. Flexible commercial work is completed within an agreed window. Commons programmes fill the gaps: for installed equipment that would otherwise sit idle, an additional workload may cost little beyond the marginal cost of electricity, so matching flexible commons work to these gaps improves utilisation and helps recover fixed costs.

Demand that supports investment

Recurring customer commitments give providers a basis for planning new equipment and facilities. The Champion brings organised demand to AI developers, specialist computing providers, major cloud platforms and equipment suppliers.

How the company coordinates work
WorkHow the company coordinates it
Live servicesCapacity matched to the customer's required quality, availability, privacy and response time.
Flexible commercial workApproved tasks completed within an agreed window, using suitable capacity when available.
Commons programmesResearch, curation and model work matched to contributed or purchased computing.

The takeawayThe commercial benefit is measured in lower cost per useful task, better service or stronger margins.

Part V · Compute, Commons and Foundation Coin

Build once. Improve together.

The commons is shared knowledge, models and software that people can use and improve. Each Champion contributes to it and begins with more than it could build alone.

The commons

Build once. Improve together.

Each Champion contributes to the commons and begins with more than it could build alone. Select a node.

Start from the commons

The commons is shared knowledge, models and software that people can use and improve. Each Champion begins with more than it could build alone.

What stays private

Private records keep their protections. Only reviewed work that can properly be shared enters the commons.

Some programmes create public goods: scientific datasets, medical evidence, local-language resources, curricula and useful models. Each deployment should leave the next company better prepared.

Figure 13

Improvements come from delivery: a tested connector, an evaluation, an adapted model or a better operating method. Private records keep their protections. Reviewed work that can properly be shared enters the commons and gives the next team a stronger starting point.

A foundation led by Champions

As the network develops, responsibility for the shared open-source software passes to a foundation led by participating Champions. II remains a contributor, research organisation and commercial provider.

Clear responsibilities across the network
ResponsibilityWhere it belongs
Local deliveryThe Champion: customers, staff, operating relationships and service obligations.
The common projectThe Champion-led foundation: shared standards, contributions, releases and maintenance.
Products and researchII and other providers: their products, commercial services and research programmes.
Equipment and financeAsset owners and lenders: equipment, financing and related obligations.

The takeawayEach deployment should leave the next company better prepared and the community better served.

Part V · Compute, Commons and Foundation Coin

Foundation Coin and common purpose

Champion equity powers local companies. Foundation Coin (FC) powers the commons, connecting participants across the network and rewarding contributions to shared knowledge, models and research.

Proof of Benefit

Foundation Coin and common purpose

Champion equity powers local companies. FC powers the commons. Select a step.

Contribute

computing and inference to accepted commons work.

Supply cap

21,000,000 FC

Rewards are distributed under the protocol rules. People may choose to hold FC because they value a network that builds common capability; providers can accept it as part of their compensation.

Figure 14

Under Proof of Benefit, AI Companies earn FC through verified contributions of computing and inference to accepted commons work. Foundation Coin has a supply cap of 21 million.

The benefit remains after the computation

The provider can earn while the resulting dataset, model or method remains available for other uses. A reviewed evidence collection can support research, professional work and future models; a local-language dataset can improve many services; a tested method can reduce the effort of another deployment.

The Champion answer

Confidence in continuing contribution

People may choose to hold FC because they value a network that builds common capability and expect its work to continue. Providers can accept it as part of their compensation.

The takeawayPublic benefit supports confidence in FC when its results are useful and its institutions behave reliably.

Part VI · Delivery and the Founding Invitation

Build accountable institutions

The council brings together people, customers and resources. Accountable leadership turns those commitments into services. Progress is established by what the company can do.

From formation to delivery

Progress is what the company can do

Five stages, bottom up. Each names the work to complete and what it takes to advance.

Stage 1

Form

Work to complete

Establish the company, founding council and core leadership.

To advance

A legally established company, named responsibilities and a funded formation plan.

An economy-wide mission begins with a small number of important customer relationships.

The test is a capable team, useful services, maintained systems and a business able to meet its commitments. Paid work must sustain delivery and support the universal service as the company grows.

Figure 15

Five stages from formation to expansion. Each names the work to complete and what it takes to advance: from a legally established company with a funded formation plan, to repeatable delivery, retained paying customers and revenue exceeding direct delivery costs.

A focused beginning

An economy-wide mission begins with a small number of important customer relationships. Before launch, the customer and Champion agree how success will be tested, the service standards and who responds when something goes wrong.

The takeawayThe test is a capable team, useful services, maintained systems and a business able to meet its commitments. Paid work must sustain delivery and support the universal service as the company grows.

Part VI · Delivery and the Founding Invitation

Help build the intelligence your home will depend on

The opportunity is to help establish a local institution while its ownership, leadership and customer relationships are being formed. The people brought together now will shape what it becomes.

The founding invitation

Bring the people and work together

The council identifies the people who can run the company, secures initial customer commitments and agrees the formation programme.

The founding council

The first objective is a capable team, committed resources, a universal agent and initial services that work.

Figure 16

A business or public leader brings a defined need; a university or profession brings talent and a training partnership; an investor or infrastructure partner brings capital, computing or equipment. Builders turn those commitments into working services.

The Champion answer

Make the first commitments concrete

The first founding discussion identifies candidate leaders, initial services and the resources participants will commit. The founding documents establish ownership, council responsibilities and investment rights.

Every community will need useful intelligence. The Champion gives local people a way to organise it, share in its economic value and have a say in its direction.
Founding Thesis, p. 17

The takeawayFound the institution. Give your community a stake in the intelligence it will depend on.

The founding invitation

Found the institution. Give your community a stake in the intelligence it will depend on.