Decision Topology for AI Enabled Software Development

Decision Topology is a proposed framework for governing consequential software work when AI can perform more of the work than traditional human-centered delivery systems were designed to permit. It treats Intent, Context, Exploration, Evidence, Judgment, Authority, Action, and Observation as a topology rather than a prescribed lifecycle. Version 0.19 reports fifteen small operational or controlled repository tests, including an executable Authority Envelope that permitted an in-authority candidate to act without redundant human approval and stopped an out-of-authority candidate until candidate-bound authorization was granted and current conditions were re-resolved. This release is a public preprint. External review is in progress. Its evidence supports bounded operational behavior in small synthetic or controlled repositories. It does not establish production effectiveness, independent assurance, organizational adoption, broad transferability, or reduced confidence cost. The evidence package retains the earlier working name AI First Software Development to preserve the identity and checksum of the artifact distributed for review.