What question did this study set out to answer?

This research aims to introduce a governance framework that allows agent systems to understand their context and human users better.

May 27, 2026Open Access

CI-01: A Constitutional Identity Layer for Host-Aware Agent Systems — Constitutional Identity Research Program, Paper 1 of 8

Key Points

This research aims to introduce a governance framework that allows agent systems to understand their context and human users better.
Formative pilot evaluation with N=15 subjects performing tasks using both baseline and CIL-augmented systems.
Comparison of baseline system metrics against those of the CIL-integrated approach.
Evaluation of task completion rates, user trust levels, and retention over 30 days.
Task completion improved from 76% to 88%.
User trust increased from 2.8 to 4.1 out of 5, indicating higher satisfaction.
Retention rates rose significantly from 42% to 78% after 30 days.

Abstract

Modern agent systems are increasingly capable, able to plan, remember, and write code. Yet they suffer from a fundamental limitation: they remain anonymous to their users. They do not intrinsically know who they serve, what their user values, or when a human life enters a crisis. This paper introduces the Constitutional Identity Layer (CIL), a formal governance framework that binds an autonomous agent system to its human and business hosts. The CIL is defined as an 8-tuple comprising Person, Business, Relationship, Life Domain, Operating Context, Permission Set, Transition State, and Emergency Policy, governed by well-defined operators for projection, scoping, delegation, escalation, and evolution. Unlike static user profiles, the CIL supports dynamic authority that shifts when roles change, life-domain separation, and explicit constitutional constraints that no agent may override. We situate this framework within the broader discourse on constitutional AI, agent alignment, and context-aware multi-agent systems, integrating it with an existing agent ecosystem. A formative pilot evaluation (N=15) comparing a baseline system to the CIL-augmented system demonstrates promising directional signals, including improvements in task completion (76% to 88%), trust (2.8 to 4.1 out of 5), and 30-day retention (42% to 78%). The CIL provides a necessary theoretical and practical foundation for host-aware agent operating systems, representing a step toward AI that is accountable, adaptive, and constitutionally aligned with human hosts. --- This is Paper 1 of 8 in the Constitutional Identity (CI) Research Program. The Constitutional Identity Research Program is an 8-paper series formalising the governance, psychology, and safety architecture of host-aware agent operating systems. The series proceeds as follows: • CI-01 (this paper): The Constitutional Identity Layer — Foundational 8-tuple, operators, and architectural integration --- Published by Pietarien (Pretoria, South Africa) and the Nexus Sovereign Project. Pietarien is building the Nexus Ecosystem — an all-round personal and business operating system powered by autonomous agents, constitutional identity, and sovereign memory architecture. Learn more: https://pietarien.com Correspondence: david@pietarien.comORCID: https://orcid.org/0009-0000-3053-6613

CI-01: A Constitutional Identity Layer for Host-Aware Agent Systems — Constitutional Identity Research Program, Paper 1 of 8

Key Points

Abstract

Cite This Study

Also Consider

Also Consider