Accountability Continuity in Persistent Multi-Agent AI Workforces: A Reference Architecture for Governed, Durable and Verifiable Delegated Work

This preprint presents a deployment-neutral reference architecture for preserving accountability across persistent multi-agent AI workforces. The paper addresses a central systems-engineering problem: how responsibility, bounded authority, execution state and verifiable evidence can remain connected as work is decomposed, delegated, transferred, executed, interrupted, resumed and accepted. The proposed architecture introduces a lifecycle-oriented design built around explicit work packages, validated task decomposition, bounded delegation, controlled execution, durable progress and recovery, evidence-based verification, conservative parent-level aggregation and auditable decision records. The manuscript identifies ten engineering requirements, defines the minimum contents of a transferable work package, describes the responsibilities and decision boundaries of the main architectural components, and analyses failure scenarios involving unauthorized delegation, stale approvals, partial execution, duplicate effects, missing evidence and premature completion. It also proposes an evaluation protocol covering ten scenario groups and measurable outcomes, including unauthorized effect rate, false completion rate, delegation integrity, evidence completeness, duplicate side-effect rate and recovery correctness. Research contribution: The paper provides a conceptual synthesis connecting established mechanisms across the full lifecycle of delegated multi-agent work. It does not claim that these individual mechanisms are new, nor does it claim empirical validation of the proposed architecture. Research methodology: Selective narrative literature review and conceptual systems-engineering analysis. Research status: Research preprint, Version 1.0. Not peer reviewed. No empirical results are reported. The architecture is deployment-neutral and is not tied to a particular software product, programming language, model provider or orchestration framework. Author: Chandra HothaORCID: 0009-0008-6353-0988

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-16
DOI
https://doi.org/10.5281/zenodo.24264516
Primary Topic
Access Control and Trust
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Accountability Continuity in Persistent Multi-Agent AI Workforces: A Reference Architecture for Governed, Durable and Verifiable Delegated Work

Chandra Hotha
Zenodo (CERN European Organization for Nuclear Research)
Access Control and Trust
preprint

Accountability Continuity in Persistent Multi-Agent AI Workforces: A Reference Architecture for Governed, Durable and Verifiable Delegated Work

Chandra Hotha
preprint en

Abstract

This preprint presents a deployment-neutral reference architecture for preserving accountability across persistent multi-agent AI workforces. The paper addresses a central systems-engineering problem: how responsibility, bounded authority, execution state and verifiable evidence can remain connected as work is decomposed, delegated, transferred, executed, interrupted, resumed and accepted. The proposed architecture introduces a lifecycle-oriented design built around explicit work packages, validated task decomposition, bounded delegation, controlled execution, durable progress and recovery, evidence-based verification, conservative parent-level aggregation and auditable decision records. The manuscript identifies ten engineering requirements, defines the minimum contents of a transferable work package, describes the responsibilities and decision boundaries of the main architectural components, and analyses failure scenarios involving unauthorized delegation, stale approvals, partial execution, duplicate effects, missing evidence and premature completion. It also proposes an evaluation protocol covering ten scenario groups and measurable outcomes, including unauthorized effect rate, false completion rate, delegation integrity, evidence completeness, duplicate side-effect rate and recovery correctness. Research contribution: The paper provides a conceptual synthesis connecting established mechanisms across the full lifecycle of delegated multi-agent work. It does not claim that these individual mechanisms are new, nor does it claim empirical validation of the proposed architecture. Research methodology: Selective narrative literature review and conceptual systems-engineering analysis. Research status: Research preprint, Version 1.0. Not peer reviewed. No empirical results are reported. The architecture is deployment-neutral and is not tied to a particular software product, programming language, model provider or orchestration framework. Author: Chandra HothaORCID: 0009-0008-6353-0988

Zenodo (CERN European Organization for Nuclear Research)
Access Control and Trust
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon
Accountability Continuity in Persistent Multi-Agent AI Workforces: A Reference Architecture for Governed, Durable and Verifiable Delegated Work — Chandra Hotha · Zenodo (CERN European Organization for Nuclear Research) (2026) | TGRS Research Map | TGRS