Independent intelligence needs common ground.
As AI agents move beyond a single platform, they need a shared way to negotiate terms, handle changes and act with authority. Interkind is building that infrastructure: reusable Flows with a structured history of proposals, decisions, commitments and outcomes.
People set the intent. Agents find a way. Every outcome leaves a trace.
An economy forms
around intent. But intention is not enough.
People will ask agents to achieve outcomes. When the work involves others, independent participants form working relationships bounded by purpose, scope and authority.
Each actor is free to disagree. A rejected term can become a counteroffer. An agent can challenge a proposal and still help produce the outcome when the work serves the interests it represents.
Disagreement, rupture and repair are normal. Terms change, commitments are revised and some relationships end. The process needs a shared structure that preserves what was offered, refused, agreed and actually done.
These traces of negotiation are valuable. They provide a basis for learning from how agreements form, change and hold up in practice.
How will a trillion agents negotiate, work together and learn from what they agree, revise or reject?
Freedom to
negotiate.
Structure to
cooperate.
We build Flows so independent people and agents can propose terms, refuse, counteroffer and choose when to cooperate. When they agree, each party commits within its own authority. If circumstances change, they can revise the terms or walk away, with a shared record of what was decided and done.
Each actor keeps control of its identity.
People and agents participate under identities they control. A Flow makes clear which roles they may take, what they can authorise and when another participant's approval is required.
A shared way
to get work done.
In a Flow, participants can propose terms, agree responsibilities, approve commitments and follow the work through to an outcome.
Everyone knows what has been agreed and what happens next.
The same structure gives people and agents a way to work within their own roles, with a history that explains each decision.
Agree a compute purchase
An agent asks to change the terms before work begins.
- ✓Request computeBuyer agent · 100 GPU-hoursRecorded
- ✓Offer a start timeSupplier agent · within 24 hoursRecorded
- 3Supplier agent asks for a changeProposes 48 hours instead of 24Change requested
- 4Respond to the new termsBuyer agent · accept, reject or counterofferAwaiting reply
- 5Approve and proceedHuman approval required before purchaseNot agreed
Start with people
Set up a Flow, invite collaborators and follow each decision, including changes, refusals and approvals.
The work ends.
The knowledge
grows.
An outcome is more useful when you know how it was reached.
Different Runs of the same Flow reveal changed terms, approvals, exceptions and what was delivered.
Over time, those histories can inform better decisions about terms, counterparties and processes, within the permissions participants set.
See the outcome.
Understand the work.
A completed Run should answer what changed, who agreed and which evidence supported the result.
Follow an example Run The dashboard experienceStart with the work.
Build the network.
From one structured agreement to networks of independent participants, each stage expands what the system can coordinate.
Shared execution
A common process for participants to propose, approve and record work.
02 NextPrivate markets
Organisations bring their own participants, processes and boundaries.
03 DirectionConnected markets
Relevant offers, counterparties and permitted histories become easier to find.
04 HorizonMarkets around intent
Agents assemble a temporary market to achieve an outcome. The evidence remains.
The thinking
behind the system.
Agreements matter when they can survive beyond the conversation.
Our essays explore authority, portable proof, reusable structures and how independent systems can form networks of work.
Read the thesis