Judgment Latency in Regulated Enterprises: A Conceptual Framework and Research Agenda for Role-Specific AI Decision Support
Roman Bodnarchuk, Co-Founder and Chief Executive Officer, WisdomTwin, Inc. Toronto, Ontario, Canada.
Working paper and preprint. Not peer reviewed. Proposed outcomes require empirical validation.
- Paper ID
- WT-100-001
- Version
- 2.0
- Publication date
- September 3, 2026
- Type and status
- Working paper / preprint. Not peer reviewed.
- DOI
- Pending repository deposit. No DOI has been issued.
- License
- CC BY 4.0
- Language
- English
- Author
- Roman Bodnarchuk
- ORCID
- 0009-0004-3113-2118
License: Creative Commons Attribution 4.0 International (CC BY 4.0). Reuse with attribution under creativecommons.org/licenses/by/4.0.
What the paper argues.
Organizations can possess the information required for a decision while still waiting for a particular role to attend to it. This conceptual working paper defines judgment latency as the elapsed time between decision readiness and attention by the person or forum whose judgment is required. It proposes the AI Judgment Twin as a role-specific, evidence-grounded decision-support architecture, distinguishes the concept from personal simulation, and derives governance requirements for provenance, authorization, safe decline, human confirmation, monitoring, and contestability. The paper integrates research on work interruption, meetings, decision rights, organizational memory, digital and human twins, and human reliance on automation. It reproduces the assumptions behind McKinsey's 2019 estimate of about 530,000 manager-days and USD 250 million in annual wages, labels the “about one day per week” restatement as author-derived arithmetic, introduces a non-overlapping model of potentially recovered time, and proposes six falsifiable propositions and a staged empirical evaluation protocol. The proposed outcomes require empirical validation.
Read this before citing.
These disclosures appear in the paper and govern how it may be represented.
- This is a conceptual working paper and preprint, not an empirical study and not peer reviewed.
- The proposed outcomes require empirical validation.
- The demonstrations referenced are founder-built product demonstrations.
- Roman Bodnarchuk is Co-Founder and CEO of WisdomTwin, Inc., is involved in developing the proposed architecture, and has a financial interest in its adoption.
- Generative AI assisted with source discovery, fact-checking, drafting, editing, and document production. The named author is responsible for the final text.
- No new empirical dataset was collected or analyzed.
- No external research funding was reported for preparation of the working paper.
How this record has changed.
Version 1.0
Initial pre-publication draft; references flagged as unverified.
Version 2.0
Fact-checked conceptual revision. Source caveats corrected, an unverifiable draft citation removed, McKinsey arithmetic reproduced and relabeled as author-derived, architecture claims converted to design requirements, propositions and evaluation protocol added, legal scope narrowed.
Version 2.0 is the version of record.
Suggested citation.
No DOI is included because none has been issued. A DOI will be added after repository deposit.
Bodnarchuk, R. (2026). Judgment latency in regulated enterprises: A conceptual framework and research agenda for role-specific AI decision support (WisdomTwin Working Paper WT-100-001, Version 2.0). WisdomTwin, Inc.
@techreport{bodnarchuk2026judgmentlatency,
author = {Bodnarchuk, Roman},
title = {Judgment Latency in Regulated Enterprises: A Conceptual Framework and Research Agenda for Role-Specific AI Decision Support},
institution = {WisdomTwin, Inc.},
type = {WisdomTwin Working Paper},
number = {WT-100-001},
year = {2026},
month = sep,
address = {Toronto, Ontario, Canada},
url = {https://wisdomtwin.ai/research/wt-100-001},
note = {Version 2.0. Working paper; not peer reviewed. DOI pending repository deposit. Record: https://wisdomtwin.ai/research/wt-100-001}
}TY - RPRT AU - Bodnarchuk, Roman TI - Judgment Latency in Regulated Enterprises: A Conceptual Framework and Research Agenda for Role-Specific AI Decision Support PY - 2026 DA - 2026/09/03 PB - WisdomTwin, Inc. CY - Toronto, Ontario, Canada ET - Version 2.0 VL - WT-100-001 LA - en KW - judgment latency KW - AI-assisted decision making KW - organizational memory KW - decision rights KW - regulated enterprises KW - human oversight KW - provenance KW - asynchronous management KW - model risk KW - human-computer interaction UR - https://wisdomtwin.ai/research/wt-100-001 UR - https://wisdomtwin.ai/research/wt-100-001 N1 - Working paper; not peer reviewed. DOI pending repository deposit. ER -
Stable records for this paper.
Public, versioned and directly accessible without login or interaction gates.
Make enterprise decisions at AI-native speed.
AI Judgment Twins for regulated enterprises: a sovereign judgment layer so high-consequence decisions move while the question is live, with a named human still accountable.
Watch a Judgment Twin handle a live enterprise decision.