1. The ObliNet protocol
A formal description of participants, interests, capabilities, commitments, authority, dependencies, results and procedures for changing rules.
ObliNet and AI: what is the fundamental difference?
Can AI organise cooperation under ObliNet’s rules on its own? Would a separate infrastructure still be needed?
ObliNet is not a particular artificial intelligence program.
We are designing an infrastructure for cooperation in which independent people and organisations can align interests, undertake interconnected commitments, achieve shared results and improve the rules of their interaction.
AI could help perform many functions of this infrastructure. Yet its intellectual capabilities alone do not mean that participants have adopted shared rules or recognised the corresponding commitments.
The central distinction: ObliNet defines how cooperation is organised, while AI can help put it into practice.
What is the difference?
| Criterion | ObliNet | AI working under ObliNet’s rules |
|---|---|---|
| Nature | A protocol and a proposed infrastructure for cooperation | An intelligent assistant, executor or operator |
| Purpose | Enable independent participants to coordinate their actions | Help apply adopted rules |
| Commitments | Arise from agreements and recognised procedures | AI helps formulate, record and verify commitments |
| Authority | Defined by participants and existing institutions | AI acts within the authority delegated to it |
| Changing rules | Provides a procedure for joint revision | AI analyses and suggests improvements |
| Shared result | Arises through participants’ coordinated actions | AI supports the organisation and verification of those actions |
One AI following rules does not make them operative rules of cooperation between people.
A simple example
Three independent participants — a client, a supplier and a contractor — want to carry out a shared project.
Option A. AI assists the participants
AI identifies interests, proposes a plan, detects contradictions and reminds people of commitments.
But if the parties have not agreed how to change terms, verify results and allocate responsibility, intelligence alone cannot fill these gaps in their arrangements.
Option B. Participants agree to work under the ObliNet protocol
They establish:
- who is responsible for what;
- which results and conditions require confirmation;
- which commitments depend on others;
- how significant changes are agreed;
- who has authority to approve decisions;
- how results are verified;
- how the rules themselves can be improved.
AI helps apply these agreements.
This changes the way cooperation is organised as well as the information available to participants.
What if AI performs all of ObliNet’s functions?
Suppose AI can:
- identify opportunities for cooperation;
- analyse participants’ interests;
- prepare mutually acceptable proposals;
- record commitments and dependencies;
- verify fulfilment;
- identify shortcomings in current rules;
- propose and support improvements to those rules.
If participants recognise the procedures, undertake commitments themselves and AI acts within defined limits of authority, such a system could become a practical implementation of the ObliNet protocol.
A separate universal platform would then not be necessary.
ObliNet could be implemented through different applications, AI agents and existing information systems that interact through a compatible protocol.
Three elements still need to be distinguished:
- Protocol
- defines the rules of cooperation.
- AI
- helps apply and improve those rules.
- Infrastructure
- supports their practical application, interoperability, record keeping, confirmation of decisions and interaction between participants.
These three elements are connected, but are not interchangeable. One AI does not automatically provide reliability, record storage, participant identification or verification of authority. These functions must be defined and provided separately.
An AI agent may act on a delegated instruction. Its authority must be defined, and responsibility addressed by the applicable agreements and rules. An AI recommendation does not itself create a commitment or authorise changes to another party’s agreement.
The main task: improving cooperation itself
AI can help complete tasks within existing arrangements. ObliNet also envisages a mechanism through which participants can:
- Identify recurring problems and new opportunities.
- Analyse which rules affect outcomes.
- Propose changes.
- Agree them with the parties concerned.
- Test alternative ways of working together.
- Approve and apply more suitable rules.
- Check their consequences and revise them when needed.
AI can support this entire process too. The project still needs to define the requirements for that support and test it in practice.
The distinction lies in a recognised mechanism that allows independent participants to change how their cooperation is organised, rather than in an intellectual ability to analyse rules.
What are we developing?
2. Intelligent tools
AI agents and other software tools that help people apply the protocol, analyse situations and prepare decisions. Different organisations could develop them; they need not belong to one platform.
3. Cooperation in practice
Real participants apply recognised rules, undertake and fulfil commitments, assess results and improve their interaction. The practical value of the proposed approach must be tested at this level.
A working AI demonstration does not establish the effectiveness of a new way of organising cooperation. The protocol and proposed ways of applying it are under development; their practical value remains to be tested.
The central research question
What must the ObliNet protocol define so that different, sufficiently capable AI systems, people and organisations can organise cooperation under shared rules — without requiring a special AI or a single centralised platform?
We hypothesise that the most important outcome of developing ObliNet could be a reproducible protocol for cooperation, usable by different people, organisations and intelligent systems.
Our task is to develop that protocol and test whether applying it helps participants make better use of their capabilities, achieve shared results and improve the rules of their interaction.