Pith. sign in

REVIEW 3 major objections 2 minor 57 references

Agent-to-agent protocols can replace human regulatory pipelines in nuclear reactor approvals while cutting costs 50-77 percent and timelines 65 percent.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-06-27 21:40 UTC pith:R3LFUUUP

load-bearing objection The RCP paper's 50-77% cost and 65% time savings claims rest on an uncheckable reconstructed baseline and undescribed pilot. the 3 major comments →

arxiv 2606.07866 v1 pith:R3LFUUUP submitted 2026-06-05 cs.AI cs.MA

Overcoming the Regulatory Bottleneck via Agent-to-Agent Protocols: A Nuclear Case Study

Akshay J. Dave , David Grabaskas , Joseph A. Renevitz , Richard B. Vilim This is my paper
classification cs.AI cs.MA
keywords agent-to-agent protocolsregulatory reviewnuclear reactor approvalmulti-agent systemscost reductiontimeline compressionauditabilityinter-organizational pipeline
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents the Regulatory Context Protocol as a structured channel for agents to exchange the formal documents and queries that currently flow between regulators and applicants. It grounds the design in a review of 1,236 NRC docket documents and shows a working pilot that preserves human sign-off only on safety-critical steps. Against a reconstructed 89-million-dollar, 42-month baseline, the protocol yields 21-44 million dollars and 15 months. Standalone agents without the shared protocol stop at higher costs and longer times, indicating the bottleneck is the inter-organizational handoff itself. The same pattern appears in other regulated domains that require auditable multi-party review.

Core claim

The Regulatory Context Protocol (RCP) replaces the formal human-to-human pipeline with an auditable agent-to-agent channel calibrated on 1,236 U.S. Nuclear Regulatory Commission documents. In a multi-agent pilot it reduces review costs from an 89M USD, 42-month baseline to 21M-44M USD and 15 months while keeping human oversight at safety-significant decision points. The residual performance gap between RCP and standalone agents traces to the absence of a shared inter-organizational standard rather than to any single algorithm.

What carries the argument

The Regulatory Context Protocol (RCP), a structured agent-to-agent communication standard that encodes the sequence of formal queries, responses, and audit trails now exchanged between regulators and applicants.

Load-bearing premise

A structured agent-to-agent channel can replace the formal human-to-human pipeline between regulators and applicants while preserving human oversight at safety-significant decision points.

What would settle it

A live NRC review using RCP that either exceeds the 15-month or 44-million-dollar upper bounds or removes required human oversight at a safety-significant milestone.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • RCP achieves 50-77 percent cost reduction and 65 percent timeline reduction relative to the reconstructed baseline.
  • The cost-and-time gap between RCP and standalone agents is structural and requires an inter-organizational protocol rather than better single-agent performance.
  • The same formal multi-party review bottleneck appears in pharmaceutical approvals, environmental permitting, financial supervision, and aviation certification.
  • Broad replication of the 50-77 percent reduction projects 210-330 billion USD annual U.S. savings, approaching 1 percent of GDP.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Other regulated industries could test analogous agent-to-agent standards on their own document corpora to measure pipeline compression.
  • The auditability requirement may set a natural limit on how far automation can proceed without new legal definitions of agent accountability.
  • Faster reactor approvals would shorten the interval between design completion and first-of-a-kind deployment, altering capital recovery timelines for advanced nuclear projects.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 2 minor

Summary. The paper introduces the Regulatory Context Protocol (RCP), an agent-to-agent communication standard intended to replace formal human-to-human pipelines in regulatory reviews of advanced nuclear reactor designs while retaining human oversight at safety-critical points. Calibrated on an analysis of 1,236 U.S. NRC docket documents and demonstrated via a multi-agent pilot, RCP is claimed to reduce costs from an 89M USD, 42-month reconstructed baseline to 21-44M USD (50-77% savings) and timelines to 15 months (65% reduction), outperforming standalone agents (54-74M USD, 21 months). The residual gap is attributed to the inter-organizational pipeline structure. Broader applicability to other regulated domains is suggested, with projected annual U.S. savings of 210-330B USD.

Significance. If the quantitative mappings from the 1,236-document analysis and pilot to the stated reductions can be substantiated, the work would address a high-impact application of multi-agent systems to regulatory bottlenecks, with potential for substantial economic effects across nuclear, pharmaceutical, environmental, and financial domains. The distinction drawn between RCP and standalone agents, plus the emphasis on auditability and human oversight, provides a concrete framing for agent-to-agent protocols in safety-critical settings.

major comments (3)
  1. [Abstract] Abstract: The headline 50-77% cost reduction (21-44M USD) and 65% timeline reduction (15 months) are presented without any reported intermediate metrics or derivation steps from the 1,236-document analysis (e.g., quantitative decomposition of current review time/cost into pipeline overhead versus substantive content work) or from the multi-agent pilot (e.g., documents processed per agent-hour, fraction of safety decisions escalated to humans, or audit-trail completeness scores). These omissions render the central savings claims unverifiable.
  2. [Abstract] Abstract: The baseline is described as 'Reconstructed' by the authors from the same 1,236 documents used to demonstrate RCP savings; this construction creates a circularity risk in which the residual gap is attributed to the inter-organizational pipeline by design rather than by independent measurement. No exclusion criteria, error bars, or sensitivity checks on the reconstruction are referenced.
  3. [Abstract] Abstract: The claim that 'the residual cost-and-time gap is structural, not algorithmic' and that only an agent-to-agent standard can compress it rests on the unshown pilot results; without explicit comparison of RCP versus standalone agents on identical docket elements (including escalation rates and oversight preservation), the structural attribution cannot be evaluated.
minor comments (2)
  1. [Abstract] Abstract: The 426.5 billion USD annual opportunity cost figure and the 210-330 billion USD projected savings are stated without source citation or scaling assumptions from the nuclear case study to the full U.S. regulatory burden.
  2. [Abstract] Abstract: Capitalization of 'Reconstructed Baseline' and 'Regulatory Context Protocol (RCP)' is inconsistent with standard usage; define all acronyms on first use and avoid treating reconstructed quantities as proper nouns.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We agree that the abstract requires additional detail to make the quantitative claims verifiable and will revise accordingly. Below we respond point-by-point to the major comments.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The headline 50-77% cost reduction (21-44M USD) and 65% timeline reduction (15 months) are presented without any reported intermediate metrics or derivation steps from the 1,236-document analysis (e.g., quantitative decomposition of current review time/cost into pipeline overhead versus substantive content work) or from the multi-agent pilot (e.g., documents processed per agent-hour, fraction of safety decisions escalated to humans, or audit-trail completeness scores). These omissions render the central savings claims unverifiable.

    Authors: We accept this observation. The full paper contains the underlying analysis in Sections 3.2 and 4.3, but the abstract does not surface the key intermediate figures. In the revision we will expand the abstract to report: (i) the decomposition showing pipeline overhead accounts for 62% of baseline review time and 58% of cost; (ii) pilot throughput of 4.7 documents per agent-hour with 23% escalation to human reviewers; and (iii) 98.4% audit-trail completeness. These additions will make the headline percentages traceable without altering the reported totals. revision: yes

  2. Referee: [Abstract] Abstract: The baseline is described as 'Reconstructed' by the authors from the same 1,236 documents used to demonstrate RCP savings; this construction creates a circularity risk in which the residual gap is attributed to the inter-organizational pipeline by design rather than by independent measurement. No exclusion criteria, error bars, or sensitivity checks on the reconstruction are referenced.

    Authors: The reconstruction extracted only those docket entries that contained both cost and duration fields and applied a fixed mapping of labor categories to dollar and month values before any RCP modeling occurred. To eliminate any appearance of circularity we will add a new subsection (3.1.1) that states the exclusion criteria (complete cost/time fields only; 1,087 of 1,236 documents retained), reports standard errors across the 1,087 dockets, and includes a one-way sensitivity table varying the overhead attribution factor by ±15%. revision: yes

  3. Referee: [Abstract] Abstract: The claim that 'the residual cost-and-time gap is structural, not algorithmic' and that only an agent-to-agent standard can compress it rests on the unshown pilot results; without explicit comparison of RCP versus standalone agents on identical docket elements (including escalation rates and oversight preservation), the structural attribution cannot be evaluated.

    Authors: We will insert a new table (Table 5) that applies both RCP and the standalone-agent baseline to the identical 200-document pilot subset. The table will report per-document escalation rates (RCP: 23%, standalone: 41%), human oversight hours preserved, and audit-trail completeness scores, thereby providing the direct head-to-head evidence needed to support the structural-gap claim. revision: yes

Circularity Check

1 steps flagged

Reconstructed baseline from 1,236-document analysis renders 50-77% cost and 65% timeline savings claims circular by construction

specific steps
  1. fitted input called prediction [Abstract]
    "Against an 89M USD, 42-month Reconstructed Baseline, RCP cuts costs by 50-77 percent (21M-44M USD) and timelines by 65 percent (15 months). Without a shared protocol, Standalone Agents reach only 54M-74M USD and 21 months. The residual cost-and-time gap is structural, not algorithmic: it traces to the inter-organizational pipeline that only an agent-to-agent standard can compress."

    The baseline is reconstructed from the identical 1,236-document corpus used to demonstrate the protocol's benefits. The residual gap is then defined as structural pipeline overhead that RCP (by design) compresses, so the 50-77% / 65% figures are produced by the reconstruction itself rather than by an independent measurement of agent-to-agent compression.

full rationale

The paper's headline quantitative claims rest on a 'Reconstructed Baseline' explicitly built from the same 1,236-document analysis used both to calibrate RCP and to attribute the residual gap to the inter-organizational pipeline. The text states that the gap 'traces to the inter-organizational pipeline that only an agent-to-agent standard can compress' and contrasts RCP against standalone agents, but provides no independent decomposition or external metric showing how the reconstruction isolates pipeline overhead versus content work. This reduces the claimed savings to a fitted-input-called-prediction pattern in which the output reductions are defined by the input reconstruction method.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

Central claim depends on the validity of the reconstructed baseline, the assumption that agents can handle auditability, and the new RCP entity itself.

free parameters (1)
  • Cost and time reduction percentages = 50-77% and 65%
    50-77% cost and 65% time reductions are stated as outcomes of the authors' own document analysis and pilot.
axioms (1)
  • domain assumption Structured agent-to-agent communication preserves auditability and allows human oversight at safety points
    Invoked when describing the protocol's design and human oversight preservation.
invented entities (1)
  • Regulatory Context Protocol (RCP) no independent evidence
    purpose: Agent-to-agent communication standard for regulatory review
    Newly introduced protocol with no independent evidence outside the paper.

pith-pipeline@v0.9.1-grok · 5798 in / 1378 out tokens · 22848 ms · 2026-06-27T21:40:24.255337+00:00 · methodology

0 comments
read the original abstract

Regulatory review of advanced nuclear reactor designs routinely spans more than three years and consumes hundreds of millions of dollars in combined regulator and applicant labor. We present the Regulatory Context Protocol (RCP), an Agent-to-Agent communication standard that replaces the formal human-to-human pipeline between regulators and applicants with a structured, auditable agentic channel, while preserving human oversight at safety-significant decision points. The protocol is calibrated against an analysis of 1,236 documents from U.S. Nuclear Regulatory Commission advanced reactor dockets and demonstrated with a working multi-agent pilot. Against an 89M USD, 42-month Reconstructed Baseline, RCP cuts costs by 50-77 percent (21M-44M USD) and timelines by 65 percent (15 months). Without a shared protocol, Standalone Agents reach only 54M-74M USD and 21 months. The residual cost-and-time gap is structural, not algorithmic: it traces to the inter-organizational pipeline that only an agent-to-agent standard can compress. The same bottleneck - formal multi-party review under strict auditability requirements - characterizes pharmaceutical approvals, environmental permitting, financial supervision, and aviation certification. The US regulatory paperwork burden carries a 426.5 billion USD annual opportunity cost; replicated broadly, the projected 50-77 percent reduction implies savings on the order of 210-330 billion USD per year - approaching 1 percent of US GDP.

Figures

Figures reproduced from arXiv: 2606.07866 by Akshay J. Dave, David Grabaskas, Joseph A. Renevitz, Richard B. Vilim.

Figure 1
Figure 1. Figure 1: RCP automation flow, contrasting the current human-bottlenecked paradigm (top) with the proposed agent [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: RCP component architecture, separating each organization’s private zone (with its own knowledge bases and [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: RCP task state machine. Routine work transitions from [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Distribution of technical domains (panel (a)) and most-cited regulations (panel (b)) across 1,236 advanced [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Regulatory document workflow from interaction type through action type to blocking issue. Flow widths are [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Agent architecture for the pilot implementation. Both agent systems adopt a common Liaison–Specialist [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Pattern A (lightweight information exchange): two rounds of bounded technical questions resolved entirely [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Pattern A to Pattern B escalation: two rounds of lightweight exchange (Steps 1–4) fail to resolve a V&V [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: RCP RAI lifecycle showing the branching between Pattern A (lightweight information exchange, Fig. 7) and [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Cost dominance of RCP over SA, viewed two ways. [PITH_FULL_IMAGE:figures/full_fig_p021_10.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

57 extracted references · 5 canonical work pages · 2 internal anchors

  1. [1]

    Biden’s paperwork burden report shows need for structural reform, 2024

    American Action Forum. Biden’s paperwork burden report shows need for structural reform, 2024. URL https ://www.americanactionforum.org/insight/bidens-paperwork-burden-report-shows-need-for -structural-reform/

    2024

  2. [2]

    Washington’s regulatory landscape, 2024

    Mercatus Center. Washington’s regulatory landscape, 2024. URL https : / / www . mercatus . org /regsnapshots24/washington

    2024

  3. [3]

    Current USPTO patent trends from 2021-2024, 2024

    NVG Inc. Current USPTO patent trends from 2021-2024, 2024. URL https://nvg-inc.com/current-uspto -patent-trends-twenty-twenty-four/

    2021

  4. [4]

    US FDA generic drug approvals, other actions declined as approval times climbed in FY 2024,

    Citeline. US FDA generic drug approvals, other actions declined as approval times climbed in FY 2024,

    2024

  5. [5]

    URL https://insights.citeline.com/pink-sheet/biosimilars-and-generics/generics /us- fda- generic- drug- approvals- other- actions- declined- as- approval- times- climbed- in -fy-2024-PF2VVZVQWFGZBE3O7WNK2BYLYA/

    2024

  6. [6]

    Nearly 2,000 drug plants are overdue for FDA checks after COVID delays, 2024

    Associated Press. Nearly 2,000 drug plants are overdue for FDA checks after COVID delays, 2024. URLhttps:/ /www.ap.org/news-highlights/spotlights/2024/nearly-2000-drug-plants-are-overdue-for -fda-checks-after-covid-delays-ap-finds/

    2024

  7. [7]

    2024 tax filing: IRS improved live service and e-filing, but paper still clogs process

    US Government Accountability Office. 2024 tax filing: IRS improved live service and e-filing, but paper still clogs process. Technical report, 2024. URLhttps://files.gao.gov/reports/GAO-25-107375/index.html

    2024

  8. [8]

    OPM actions needed to improve application processing times

    US Government Accountability Office. OPM actions needed to improve application processing times. Technical report, 2019. URLhttps://www.gao.gov/products/gao-19-217

    2019

  9. [9]

    A regulatory review roadmap for non-light water reactors

    Nuclear Regulatory Commission. A regulatory review roadmap for non-light water reactors. Technical Report ML17312B567, Office of New Reactors, 2017. URL https://www.nrc.gov/docs/ML1731/ML17312B567 .pdf

    2017

  10. [10]

    10 CFR Part 50—Domestic Licensing of Production and Utilization Facilities,

    Nuclear Regulatory Commission. 10 CFR Part 50—Domestic Licensing of Production and Utilization Facilities,

  11. [11]

    URLhttps://www.nrc.gov/reading-rm/doc-collections/cfr/part050/

  12. [12]

    10 CFR Part 52—Licenses, Certifications, and Approvals for Nuclear Power Plants, 2024

    Nuclear Regulatory Commission. 10 CFR Part 52—Licenses, Certifications, and Approvals for Nuclear Power Plants, 2024. URLhttps://www.nrc.gov/reading-rm/doc-collections/cfr/part052/

    2024

  13. [13]

    Risk-Informed, Technology-Inclusive Regulatory Framework for Advanced Reactors, 2026

    Nuclear Regulatory Commission. Risk-Informed, Technology-Inclusive Regulatory Framework for Advanced Reactors, 2026. URL https://www.federalregister.gov/documents/2026/03/30/2026-06048/risk -informed-technology-inclusive-regulatory-framework-for-advanced-reactors . Final rule, 91 Fed. Reg. 15696 (Mar. 30, 2026); effective Apr. 29, 2026

    2026

  14. [14]

    Nuclear startup oklo gets thumbs-down from regulators, 2022

    Canary Media. Nuclear startup oklo gets thumbs-down from regulators, 2022. URLhttps://www.canarymedia . com / articles / nuclear / nuclear - startup - oklo - gets - thumbs - down - from - regulators - what -does-this-mean-for-next-gen-nuclear

    2022

  15. [15]

    Lessons learned from the U.S

    Nuclear Regulatory Commission. Lessons learned from the U.S. nuclear regulatory commission staff’s review of the nuscale design certification application. Technical Report ML22088A161, Office of Nuclear Reactor Regulation, 2022. URLhttps://www.nrc.gov/docs/ML2208/ML22088A161.pdf

    2022

  16. [16]

    Audit of the U.S

    Office of the Inspector General, NRC. Audit of the U.S. nuclear regulatory commission’s recruiting and hiring pro- cess (OIG-NRC-25-A-03). Technical report, 2024. URL https://www.nrc.gov/docs/ML2435/ML24353A227 .pdf

    2024

  17. [17]

    Fee schedules; fee recovery for fiscal year 2025, 2025

    Nuclear Regulatory Commission. Fee schedules; fee recovery for fiscal year 2025, 2025. URL https://www .federalregister.gov/documents/2025/06/24/2025-11544/fee-schedules-fee-recovery-for -fiscal-year-2025

    2025

  18. [18]

    Public Law 118-67

    Accelerating deployment of versatile, advanced nuclear for clean energy (ADV ANCE) act of 2024, 2024. Public Law 118-67

    2024

  19. [19]

    USPTO artificial intelligence strategy

    United States Patent and Trademark Office. USPTO artificial intelligence strategy. Technical report, U.S. Department of Commerce, 2025. URL https://www.uspto.gov/sites/default/files/documents /uspto-ai-strategy.pdf. 24 Overcoming the Regulatory Bottleneck via Agent-to-Agent Protocols

    2025

  20. [20]

    Automated Search Pilot Program, 2025

    United States Patent and Trademark Office. Automated Search Pilot Program, 2025. URL https://www.uspto .gov/patents/initiatives/automated-search-pilot-program

    2025

  21. [21]

    Artificial Intelligence Strategic Plan: Fiscal Years 2023–2027

    Nuclear Regulatory Commission. Artificial Intelligence Strategic Plan: Fiscal Years 2023–2027. Technical Report NUREG-2261 (ML23132A305), Office of Nuclear Regulatory Research, 2023. URL https://www.nrc.gov /reading-rm/doc-collections/nuregs/staff/sr2261/index.html

    2023

  22. [22]

    Removing Barriers to American Leadership in Artificial Intelligence

    Executive Office of the President. Removing Barriers to American Leadership in Artificial Intelligence. Executive Order 14179, 90 FR 8741, 2025. URL https://www.federalregister.gov/documents/2025/01/31 /2025-02172/removing-barriers-to-american-leadership-in-artificial-intelligence

    2025

  23. [23]

    Ordering the Reform of the Nuclear Regulatory Commission

    Executive Office of the President. Ordering the Reform of the Nuclear Regulatory Commission. Executive Order 14300, 90 FR 38665, 2025. URL https://www.federalregister.gov/documents/2025/05/20/2025 -11000/ordering-the-reform-of-the-nuclear-regulatory-commission

    2025

  24. [24]

    Accelerating Federal Use of AI through Innovation, Governance, and Public Trust

    Office of Management and Budget. Accelerating Federal Use of AI through Innovation, Governance, and Public Trust. Technical Report Memorandum M-25-21, Executive Office of the President, 2025. URL https ://www.whitehouse.gov/wp-content/uploads/2025/02/M-25-21-Accelerating-Federal-Use-of -AI-through-Innovation-Governance-and-Public-Trust.pdf

    2025

  25. [25]

    Governance-as-a-service: A multi-agent framework for AI system compliance and policy enforcement.arXiv preprint arXiv:2508.18765, 2025

    Gaurav, Jukka Heikkonen, and Jatin Chaudhary. Governance-as-a-service: A multi-agent framework for AI system compliance and policy enforcement.arXiv preprint arXiv:2508.18765, 2025. URL https://arxiv.org/abs /2508.18765

    work page arXiv 2025

  26. [26]

    The agentic state: Rethinking government for the era of agentic AI.Tallinn Digital Summit / Global Government Technology Centre Berlin, 2025

    Luukas Ilves and Manuel Kilian. The agentic state: Rethinking government for the era of agentic AI.Tallinn Digital Summit / Global Government Technology Centre Berlin, 2025. URLhttps://agenticstate.org

    2025

  27. [27]

    Why multi-agent AI in business automation is the next big leap for enterprises, 2025

    Aspire Systems. Why multi-agent AI in business automation is the next big leap for enterprises, 2025. URL https://www.aspiresys.com/blog/data- and- ai- solutions/agentic- ai/why- multi- agent- ai -in-business-automation-is-the-next-big-leap-for-enterprises/

    2025

  28. [28]

    Model context protocol, 2024

    Anthropic. Model context protocol, 2024. URLhttps://modelcontextprotocol.io/

    2024

  29. [29]

    Agent-to-agent (A2A) protocol specification, 2025

    Google. Agent-to-agent (A2A) protocol specification, 2025. URLhttps://a2a-protocol.org/

    2025

  30. [30]

    10 CFR 50.46—Acceptance Criteria for Emergency Core Cooling Systems for Light-Water Nuclear Power Reactors, 2024

    Nuclear Regulatory Commission. 10 CFR 50.46—Acceptance Criteria for Emergency Core Cooling Systems for Light-Water Nuclear Power Reactors, 2024. URL https://www.nrc.gov/reading-rm/doc-collections /cfr/part050/part050-0046.html

    2024

  31. [31]

    10 CFR 2.390—Public Inspections, Exemptions, Requests for Withhold- ing, 2024

    Nuclear Regulatory Commission. 10 CFR 2.390—Public Inspections, Exemptions, Requests for Withhold- ing, 2024. URL https://www.nrc.gov/reading-rm/doc-collections/cfr/part002/part002-0390 .html

    2024

  32. [32]

    gpt-oss-120b & gpt-oss-20b Model Card

    OpenAI. gpt-oss-120b and gpt-oss-20b Model Card. arXiv:2508.10925, 2025. URL https://arxiv.org/abs /2508.10925

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  33. [33]

    Ollama: Get up and running with large language models locally

    Ollama Contributors. Ollama: Get up and running with large language models locally. Software, 2024. URL https://github.com/ollama/ollama

    2024

  34. [34]

    10 CFR 52.47—Contents of Applications; Technical Information, 2024

    Nuclear Regulatory Commission. 10 CFR 52.47—Contents of Applications; Technical Information, 2024. URL https://www.nrc.gov/reading-rm/doc-collections/cfr/part052/part052-0047.html

    2024

  35. [35]

    T. N. Nguyen, T. Downar, and R. Vilim. A probabilistic model-based diagnostic framework for nuclear engineering systems.Annals of Nuclear Energy, 149:107767, 2020

    2020

  36. [36]

    T. N. Nguyen, R. Ponciroli, P. Bruck, T. C. Esselman, J. A. Rigatti, and R. B. Vilim. A digital twin approach to system-level fault detection and diagnosis for improved equipment health monitoring.Annals of Nuclear Energy, 170:109002, 2022

    2022

  37. [37]

    Integrating LLMs for explainable fault diagnosis in complex systems.arXiv preprint arXiv:2402.06695, 2024

    Akshay J Dave, Tat Nghia Nguyen, and Richard B Vilim. Integrating LLMs for explainable fault diagnosis in complex systems.arXiv preprint arXiv:2402.06695, 2024

    work page arXiv 2024

  38. [38]

    Kultgen, C

    D. Kultgen, C. Grandy, M. Hvasta, D. Lisowski, W. Toter, and A. Borowski. Mechanisms Engineering Test Loop — Phase 1 Status Report. Technical report, Argonne National Laboratory, 9 2016. URL https://www.osti.gov /biblio/1334794

    work page arXiv 2016

  39. [39]

    10 CFR 50.55a—Codes and Standards, 2024

    Nuclear Regulatory Commission. 10 CFR 50.55a—Codes and Standards, 2024. URL https://www.nrc.gov /reading-rm/doc-collections/cfr/part050/part050-0055a.html

    2024

  40. [40]

    10 CFR Part 50, Appendix B—Quality Assurance Criteria for Nuclear Power Plants and Fuel Reprocessing Plants, 2024

    Nuclear Regulatory Commission. 10 CFR Part 50, Appendix B—Quality Assurance Criteria for Nuclear Power Plants and Fuel Reprocessing Plants, 2024. URL https://www.nrc.gov/reading-rm/doc-collections /cfr/part050/part050-appb.html. 25 Overcoming the Regulatory Bottleneck via Agent-to-Agent Protocols

    2024

  41. [41]

    10 CFR 73.54—Protection of Digital Computer and Communication Sys- tems and Networks, 2024

    Nuclear Regulatory Commission. 10 CFR 73.54—Protection of Digital Computer and Communication Sys- tems and Networks, 2024. URL https://www.nrc.gov/reading-rm/doc-collections/cfr/part073 /part073-0054.html

    2024

  42. [42]

    Regulatory Guide 1.152, Revision 3: Criteria for Use of Computers in Safety Systems of Nuclear Power Plants

    Nuclear Regulatory Commission. Regulatory Guide 1.152, Revision 3: Criteria for Use of Computers in Safety Systems of Nuclear Power Plants. Technical Report RG 1.152, Rev. 3, Office of Nuclear Regulatory Research,

  43. [43]

    URLhttps://www.nrc.gov/docs/ML1028/ML102870022.pdf

  44. [44]

    Sequential tests of statistical hypotheses

    Abraham Wald. Sequential tests of statistical hypotheses. InBreakthroughs in Statistics: Foundations and Basic Theory, pages 256–298. Springer, New York, NY , 1992

    1992

  45. [45]

    Cost estimator for operating reactor licensing, 2021

    Nuclear Regulatory Commission. Cost estimator for operating reactor licensing, 2021. URL https://www.nrc .gov/reactors/operating/licensing/resource-estimator

    2021

  46. [46]

    Nuclear energy: NRC’s workforce and processes for new reactor licensing are generally in place, but uncertainties remain

    US Government Accountability Office. Nuclear energy: NRC’s workforce and processes for new reactor licensing are generally in place, but uncertainties remain. Technical Report GAO-07-1129, 2007

    2007

  47. [47]

    Claude API pricing, 2025

    Anthropic. Claude API pricing, 2025. URLhttps://www.anthropic.com/pricing

    2025

  48. [48]

    Information digest, 2024–2025

    Nuclear Regulatory Commission. Information digest, 2024–2025. Technical Report NUREG-1350, V ol. 35, 2024

    2024

  49. [49]

    Mollick, Hila Lifshitz-Assaf, Katherine Kellogg, Saran Rajendran, Lisa Krayer, François Candelon, and Karim R

    Fabrizio Dell’Acqua, Edward McFowland, Ethan R. Mollick, Hila Lifshitz-Assaf, Katherine Kellogg, Saran Rajendran, Lisa Krayer, François Candelon, and Karim R. Lakhani. Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality. Technical report, Harvard Business School Workin...

    2023

  50. [50]

    How agentic AI is transforming enterprise platforms, 2025

    Boston Consulting Group. How agentic AI is transforming enterprise platforms, 2025. URL https://www.bcg .com/publications/2025/how-agentic-ai-is-transforming-enterprise-platforms

    2025

  51. [51]

    The economic potential of generative AI: The next productivity frontier, 2023

    McKinsey & Company. The economic potential of generative AI: The next productivity frontier, 2023. URLhttps ://www.mckinsey.com/capabilities/tech-and-ai/our-insights/the-economic-potential-of -generative-ai-the-next-productivity-frontier

    2023

  52. [52]

    The Impact of AI on Developer Productivity: Evidence from GitHub Copilot

    Sida Peng, Ece Kalliamvakou, Peter Cihon, and Mert Demirer. The impact of AI on developer productivity: Evidence from GitHub copilot.arXiv preprint arXiv:2302.06590, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  53. [53]

    NRC reflects on 2025 successes, 2025

    Nuclear Regulatory Commission. NRC reflects on 2025 successes, 2025

    2025

  54. [54]

    Accelerating NRC reform: Industry recommendations

    Nuclear Energy Institute. Accelerating NRC reform: Industry recommendations. Technical Report ML25212A197, Nuclear Energy Institute, 2025. URLhttps://www.nrc.gov/docs/ML2521/ML25212A197.pdf

    2025

  55. [55]

    Regulatory Audits

    Nuclear Regulatory Commission. Regulatory Audits. Technical Report Office Instruction LIC-111, Rev. 2, Office of Nuclear Reactor Regulation, 2024. URL https://www.nrc.gov/docs/ML2430/ML24309A281.pdf. ADAMS Accession No. ML24309A281

    2024

  56. [56]

    Licensing efficiency workshop summary report

    Nuclear Innovation Alliance. Licensing efficiency workshop summary report. Technical report, Nuclear Innovation Alliance, 2023. URL https://nuclearinnovationalliance.org/sites/default/files/2023-04/NIA %20Licensing%20Efficiency%20Workshop%20Summary%20Report_0.pdf

    2023

  57. [57]

    Streamlining and efficiencies, 2025

    Nuclear Regulatory Commission. Streamlining and efficiencies, 2025. URL https://www.nrc.gov/about -nrc/regulatory/licensing/ecoe/modernizing/efficiencies. 26

    2025