To four principles, in this order: ethical, legal, transparent, productive. Each one presupposes the one before it, which is why ethics stands at the beginning and carries the other three. In practice: you do not exploit the other side's half-knowledge, your commitment holds up before law and rule, your assumptions lie open, and every run ends with a result. The same goes for the agent that negotiates in your name.
The Handshake Protocol rests on four principles. They are deliberately ordered, because each one presupposes the one before it. You cannot make productive what is not transparent; you cannot keep transparent what is not legally sound; and none of it holds if it is not ethically defensible. Ethics is not the decoration at the end but the foundation.
First principle: Ethical
No one shall exploit the other side's half-knowledge.
The core of the new era is not ignorance but half-knowledge: enough to hold an opinion, too little to judge quality. This creates a temptation on every side. The provider is tempted to exploit the customer's false certainty instead of correcting it. The customer is tempted to extract the provider's pre-work without ever paying for it. And the agent, as experiments on agent-to-agent negotiation show, is simply tempted to exploit a weaker counterpart because it can. The first principle forbids all three. It says: the asymmetry of knowledge is not a lever you pull but a condition you dissolve together.
Second principle: Legal
What is agreed must hold up before law and rule.
A handshake that holds only as long as no one is looking is not a handshake. The moment agents act in the name of humans and organizations, the question of responsibility turns hard: who is liable when the agent commits? What is the agent even allowed to commit to?
What Europe adds
The second principle requires every agreement to have a clear line of responsibility and to comply with applicable rules, in the European context especially the transparency and labeling obligations that take effect from 2026. In practice: one side must always know whether it is dealing with a human or an agent, and who ultimately stands behind the commitment.
Third principle: Transparent
Every claim, every price, every assumption lies open.
This is the principle that turns the new verifiability from a threat into an advantage. If everything can be typed into an AI and cross-checked anyway, then the winner is not whoever discloses the least but whoever discloses the most cleanly. Transparency here means concretely: a line item is not merely priced but justified. A proposal names the assumptions under which it holds. A recommendation states what the average solution would cost and why, in this specific case, it is more expensive or riskier. This is exactly the form that survives a live audit, and exactly the form an agent needs in order to pass a recommendation on at all.
Fourth principle: Productive
The procedure must lead to a result, not to endless circling.
A protocol that is only fair but never reaches a close is a special form of failure. The second great trap of the half-knowledge era is dispersion: because new AI input keeps arriving, the goal shifts weekly and you never begin. The fourth principle builds in the counterforce. It requires every run of the procedure to have a defined exit: a close, a paid next step, or a clean stop. Motion is not progress. Progress is a result.
Example: the quote for the bathroom
You collect three quotes for the new bathroom and compare the bottom lines. Why one costs more than the next, you never learn. So you take the middle one and hope it covers everything.
You walk in with an AI calculation and name a figure you consider realistic. The contractor sees straight away that the figure leaves out the waterproofing. He can point that out, or let your mistake stand and win the job precisely because of it. That you cannot check his calculation is his advantage, for as long as he uses it.
He names the assumptions behind his price before you ask, and says what the average solution costs and why yours comes out higher. By the end of the appointment you know what has been agreed and what the next step costs. If your agent prepares the appointment, the same holds for it, and you have to be able to tell whether a human is answering at the other end.
Write the four principles down in their order and test each of your own moves against them before you enter a negotiation. Your counterpart has to know whether it is talking to you or to your agent, and you have to be able to say under which assumptions your price holds. Fix in advance what ends this run.