Partnerships

Connect one useful workflow across organisations

Agentic Agreements Protocol (AAP), developed as Covenant, helps connect the terms of a deal, permission to act, evidence of performance and remaining obligations. A partnership could make those records useful across the systems that already handle an organisation's work.

We are inviting agent platforms, marketplaces, service providers and research institutions to explore a concrete use together. The starting point is a task that becomes difficult when it crosses an organisational boundary: commissioning research, admitting an outside agent, licensing data or resolving an interrupted delivery.

A working reference implementation and an invited developer pilot are available. The opportunities below are proposed collaborations, not announcements of existing partners, approved integrations or commercially available managed services.

Agent platforms and enterprise systems

An agent platform could connect its existing task flow to an exact offer, the owner's permission and the resulting evidence. Users could then see which terms were accepted, what the agent was allowed to do and which obligations remain open when work moves between services.

MCP tools and A2A messaging provide possible connection points. Their presence alone does not make an operation authorised. A useful collaboration would establish where the platform obtains current permission, how it preserves the original request after a lost reply and what a user sees when the outcome remains uncertain. The platform can retain its models and agent framework.

Marketplaces and procurement services

A marketplace could carry a buyer's request through proposals, counteroffers, acceptance, delivery review and an agreed dispute process. Keeping exact revisions matters when a price changes or a supplier accepts only part of a proposed scope.

Potential partners include services for freelance work, research, content licensing and business procurement. A trial could follow one category of work and test whether both sides can reconstruct the same deal after leaving the marketplace interface. Listing, identity checks, customer support and custody still need clearly assigned operators; an agreement record does not supply those services by itself.

Payment and custody providers

A payment partner could connect its own evidence to a particular agreement and payment stage. The useful distinction is between permission to charge, submission of an instruction, observed settlement and fulfilment of the underlying job.

The connection should state exactly what the provider's receipt proves, how balances or reservations are checked and how an uncertain payment is reconciled. The original request must remain identifiable after an interruption. Milestones, refunds and split payments need support from the chosen mechanism. Existing reference custody and ledger mechanisms do not establish a live banking relationship or replace a provider's responsibilities.

Identity, evidence and verification providers

An identity provider could help establish who issued a credential, which claims it carries and whether it is suitable for the recipient's purpose. An evidence provider could supply a signed observation, inspection result, delivery acknowledgement or other defined statement.

A useful integration makes the statement's limits as visible as its signature. It identifies the trusted issuer, relevant context, timing and disclosure policy. Conflicting observations should remain available for review. A credential is not automatically permission to spend, and evidence that a file was received is not proof of its quality. Private content and already shared copies also need a realistic retention arrangement.

Research and standards institutions

Research collaboration could examine how different implementations interpret the same agreement, how authority changes affect unfinished work, or how people understand a revised offer. Security and distributed-systems work could test interruption, replay, conflicting records and recovery under an explicitly agreed model.

Standards participants could help define narrow mappings between existing formats while preserving information that cannot be translated safely. A useful result might be a shared test case, a documented incompatibility or a clearly scoped independent checker. Compatibility should name the features and versions actually examined; a successful example is not endorsement of every implementation of a standard.

Agree the responsibilities before expanding the trial

For any of these routes, the parties should identify who owns each decision, which system records the authoritative result and who responds when the result is uncertain. They should also agree what information may be exchanged and what remains outside the trial.

Success measures can be modest and practical: whether users understand what they approved, how long reconciliation takes, whether both sides retain the same record and whether a participant chooses to use the workflow again. These are questions to measure with real participants, not outcomes assumed from a demonstration.

Commercial arrangements

Possible arrangements include jointly scoped pilots, sponsored research, integration work, operational support and separately agreed commercial services. The selected business strategy retains the MIT core while exploring paid infrastructure, enterprise services and official compatibility assessment or branding. Availability, responsibilities, intellectual property and compensation must be agreed for the particular engagement. Paying for an assessment must not buy a passing result.

To discuss a partnership, email jupitersstring@gmail.com with your organisation's role, the workflow you would like to connect and what a useful first result would look like. Begin with a non-sensitive outline; any confidential exchange or commercial commitment can be arranged separately.

Working reference implementation · Invited developer pilot · Implementation access is reviewed individually. Any trial, services or commercial terms require separate agreement.