Manifesto 0.3: boundaries first, and the intent team
Structured intent now starts with boundary analysis, and treats structuring intent as a team effort in which agents propose and people decide.
Version 0.3 of the manifesto adds two ideas, along with new pages for leaders, a maturity model and a guide to hiring intent architects.
Boundaries come first. Before writing a single clause, decide what an intent covers and what it doesn’t, which decisions the software makes on its own, which it only recommends, which only a person may make, who rules on each part, and what the system trusts without checking. Security teams will recognize the idea from the authorization boundary that starts every accreditation. For organizations adopting AI agents, the boundary answers the most important question of all: what are we letting AI decide?
Agents propose; people decide. Agents can and should guess: drafting clauses, proposing boundaries and flagging contradictions. But every enterprise is unique. The people in its roles, from intent owners and domain experts to engineers, assurance and risk, judge the intent and represent their departments’ interests. Their rulings become the intent.