pith. sign in

arxiv: 2605.18935 · v1 · pith:O2F63W6Wnew · submitted 2026-05-18 · 💰 econ.EM

The Agentic Economy: Humans, AI Agents, Robots, and the Measurable Transition toward Distributed Economic Action

Davit Gondauri , Mikheil Batiashvili This is my paper

Pith reviewed 2026-05-20 01:41 UTC · model grok-4.3

classification 💰 econ.EM
keywords agentic economyAI agentsindustrial robotseconomic transitionaction capacitylabor reallocationcompute energy demanddistributed action
0
0 comments X

The pith

Economic action is distributing across AI agents, robots, and protocols, making classical categories of labor, capital, and firm incomplete.

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

The paper develops the concept of the agentic economy to describe a transition in which economic actions are increasingly performed by AI agents, industrial robots, executable protocols, and energy systems in addition to humans. It employs a conceptual-empirical approach using public data on AI investment and adoption, robot stocks, data centre electricity, and labour markets to create indicators of relative growth and concentration. A sympathetic reader would care if this transition is real because it implies that economic analysis must account for distributed decision-making and responsibility in ways that go beyond traditional human-centric models. The results indicate accelerating AI use, persistent robot capacity, and labour reallocation rather than elimination, supporting the need for a new action-capacity framework.

Core claim

The central claim is that the agentic economy is characterised by distributed economic action among humans, AI agents, industrial robots, protocols, compute, and energy. Public data transformed into growth multipliers, stock-flow ratios, and HHI measures show accelerating AI adoption, broad AI capital allocation, ongoing robotic cyber-physical capacity, rising electricity demands, and labour task reallocation. This demonstrates the incompleteness of classical categories and motivates an action-capacity framework that connects agent capacity, robotic capacity, compute-energy coupling, protocolisation, auditable trust, and human sovereignty. The agentic economy is not yet complete but its tr

What carries the argument

The action-capacity framework, which integrates model/software-agent capacity, robotic capacity, compute-energy coupling, protocolisation, auditable trust, and human sovereignty to analyse the distribution of economic action.

Load-bearing premise

The selected public institutional data sources and the derived indicators such as relative growth, CAGR, stock-flow ratios, and concentration measures adequately capture the distribution of economic action and show the incompleteness of classical categories.

What would settle it

A reversal in trends where AI adoption and robot installations slow significantly while labour data shows net job losses without corresponding task shifts would undermine the claim of a measurable transition to distributed economic action.

Figures

Figures reproduced from arXiv: 2605.18935 by Davit Gondauri, Mikheil Batiashvili.

Figure 5.1
Figure 5.1. Figure 5.1: AI adoption acceleration across official and survey indicators [PITH_FULL_IMAGE:figures/full_fig_p038_5_1.png] view at source ↗
Figure 5.2
Figure 5.2. Figure 5.2: Reported private AI investment by leading countries, 2024 [PITH_FULL_IMAGE:figures/full_fig_p039_5_2.png] view at source ↗
Figure 5.3
Figure 5.3. Figure 5.3: Data-centre electricity demand: observed value and IEA Base Case projection Source: Author’s visualisation based on International Energy Agency values used in the Results section. Projection values are not realised facts and depend on IEA Base Case assumptions [PITH_FULL_IMAGE:figures/full_fig_p040_5_3.png] view at source ↗
Figure 5.4
Figure 5.4. Figure 5.4: New industrial robot installations by region, 2024 Source: Author’s visualisation based on International Federation of Robotics values used in the Results section [PITH_FULL_IMAGE:figures/full_fig_p041_5_4.png] view at source ↗
Figure 5.5
Figure 5.5. Figure 5.5: Projected labour-market reallocation by 2030 Source: Author’s visualisation based on World Economic Forum values used in the Results section. Values are projections, not realised outcomes [PITH_FULL_IMAGE:figures/full_fig_p042_5_5.png] view at source ↗
read the original abstract

This article develops the concept of the agentic economy and diagnoses its measurable preconditions: a transition in which economic action is increasingly distributed among humans, AI agents, industrial robots, executable protocols, compute infrastructures, and energy systems. The paper argues that classical categories such as labour, capital, firm, market, productivity, and trust remain necessary but incomplete when technologies prepare decisions, coordinate workflows, support tasks, verify transactions, and reshape responsibility. Methodologically, the study uses a conceptual-empirical quantitative diagnostic design rather than a causal econometric model. It relies on public institutional data on AI investment, AI adoption, robot installations and operational stock, data-centre electricity demand, and labour-market reallocation. The reported values are transformed through transparent indicators such as relative growth, CAGR, growth multipliers, stock-flow ratios, concentration ratios, and HHI. The results show that AI adoption is accelerating, AI investment signals broad capital allocation, industrial robots represent persistent cyber-physical action capacity, compute expansion increases data-centre electricity pressure, and labour projections are more consistent with task reallocation than labour disappearance. The article contributes an action-capacity framework linking model/software-agent capacity, robotic capacity, compute-energy coupling, protocolisation, auditable trust, and human sovereignty. It concludes that the agentic economy is not yet a completed global order, but its transition pressure is measurable enough to require a distinct economic vocabulary, reproducible diagnostics, and future sector-level measurement.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper develops the concept of the 'agentic economy' as a transition in which economic action becomes distributed across humans, AI agents, industrial robots, executable protocols, compute infrastructures, and energy systems. It argues that classical categories (labour, capital, firm, market, productivity, trust) remain necessary but incomplete, and supports this via a conceptual-empirical quantitative diagnostic design that applies transparent transformations (CAGR, growth multipliers, stock-flow ratios, concentration ratios, HHI) to public institutional data on AI investment/adoption, robot installations/stock, data-centre electricity demand, and labour-market reallocation. The results document accelerating AI adoption, broad capital allocation to AI, persistent cyber-physical robot capacity, rising electricity pressure from compute, and task reallocation rather than labour disappearance. The contribution is an action-capacity framework linking model/software-agent capacity, robotic capacity, compute-energy coupling, protocolisation, auditable trust, and human sovereignty, concluding that the transition pressure is measurable enough to require distinct vocabulary and future sector-level measurement.

Significance. If the interpretive step from descriptive indicators to the incompleteness of classical categories is secured, the paper offers a timely synthesis that could stimulate new measurement agendas at the intersection of AI, robotics, energy, and economic organisation. The transparent use of public data and reproducible indicators is a strength, as is the explicit linkage of multiple capacity dimensions into a single framework. However, the significance is primarily conceptual and diagnostic rather than providing causal identification or falsifiable predictions that would immediately alter standard econometric practice.

major comments (2)
  1. [Methodological description of the conceptual-empirical quantitative diagnostic design] Methodological section (conceptual-empirical quantitative diagnostic design): the central claim that classical categories are incomplete and require replacement by an action-capacity framework is not secured by the reported diagnostics. The indicators document expansion, concentration, and reallocation, but supply no explicit test, counterfactual, or derivation showing why the same phenomena cannot be re-described as extensions of existing categories (e.g., embodied technical change, new capital services, or outsourced labour). Without this step, the empirical component supports only that 'things are changing' rather than that 'the categories must be replaced.'
  2. [Results and conclusion] Results and conclusion sections: the reported values (accelerating AI adoption, robot stock, electricity pressure, task reallocation) are consistent with the data transformations described, yet the leap to 'necessitate a distinct economic vocabulary' rests on interpretive framing rather than a formal argument ruling out subsumption under standard models. A dedicated subsection deriving the necessity of new categories (or showing why standard extensions are insufficient) would make the load-bearing claim proportionate to the evidence.
minor comments (2)
  1. [Data and indicators] Clarify the precise definitions and data sources for each indicator (CAGR, growth multipliers, stock-flow ratios, HHI) in a dedicated table or appendix to enhance reproducibility.
  2. [Abstract and introduction] The abstract and introduction use 'incomplete' and 'necessitate a distinct vocabulary' interchangeably; distinguish these claims more sharply to avoid conflating descriptive change with categorical replacement.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments on our manuscript. The report accurately captures the paper's diagnostic approach and its focus on measurable transition pressures. We address each major comment below, outlining our responses and the revisions we plan to implement.

read point-by-point responses

  1. Referee: Methodological section (conceptual-empirical quantitative diagnostic design): the central claim that classical categories are incomplete and require replacement by an action-capacity framework is not secured by the reported diagnostics. The indicators document expansion, concentration, and reallocation, but supply no explicit test, counterfactual, or derivation showing why the same phenomena cannot be re-described as extensions of existing categories (e.g., embodied technical change, new capital services, or outsourced labour). Without this step, the empirical component supports only that 'things are changing' rather than that 'the categories must be replaced.'

    Authors: We appreciate the referee's emphasis on securing the interpretive step from diagnostics to category incompleteness. Our methodological design is conceptual-empirical and diagnostic by intent, aiming to quantify distributed action capacities rather than to provide econometric tests or counterfactuals. The data transformations illustrate phenomena such as autonomous AI agent coordination and protocol-executed transactions that introduce new dimensions of agency and responsibility. These are not fully captured by extensions like embodied technical change, which typically model productivity effects within existing labour-capital frameworks without addressing the delegation of decision initiation to non-human agents. Nevertheless, we acknowledge that an explicit contrast would strengthen the manuscript. We will add a subsection to the methodological section that derives why standard category extensions are insufficient for the observed patterns of task reallocation and multi-agent coordination, using the reported indicators as illustrations. revision: yes

  2. Referee: Results and conclusion sections: the reported values (accelerating AI adoption, robot stock, electricity pressure, task reallocation) are consistent with the data transformations described, yet the leap to 'necessitate a distinct economic vocabulary' rests on interpretive framing rather than a formal argument ruling out subsumption under standard models. A dedicated subsection deriving the necessity of new categories (or showing why standard extensions are insufficient) would make the load-bearing claim proportionate to the evidence.

    Authors: We agree that the conclusion would benefit from a more formal derivation to support the call for distinct vocabulary. The current argument rests on the cumulative diagnostics across AI, robotics, energy, and labour domains, which collectively point to action capacities that classical categories treat as peripheral. For example, the growth in AI agents and executable protocols shifts the locus of economic initiative in ways not reducible to outsourced labour or new capital services alone. To address the referee's suggestion, we will introduce a dedicated subsection, likely in the results or as a bridge to the conclusion, that explicitly rules out subsumption by considering alternative framings and showing their limitations based on the empirical patterns. This revision will make the interpretive claims more proportionate while maintaining the paper's diagnostic focus. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation relies on external data and standard metrics

full rationale

The paper applies transparent transformations (CAGR, growth multipliers, stock-flow ratios, concentration ratios, HHI) to public institutional series on AI adoption, robot stock, data-centre electricity, and labour reallocation. These are descriptive indicators drawn from outside sources rather than parameters fitted inside the paper or quantities defined by the target claims. The action-capacity framework is presented as an interpretive contribution linking observed trends to classical-category incompleteness, but the quantitative steps do not reduce to self-definition, self-citation chains, or renaming of fitted inputs. No equations or uniqueness theorems are invoked that collapse back onto the paper's own premises. The derivation therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 2 invented entities

The central contribution rests on conceptual innovation and interpretation of existing public data rather than new data collection or formal modeling.

axioms (1)
  • domain assumption Public institutional data on AI investment, adoption, robot installations, data-centre electricity demand, and labour-market reallocation are accurate and representative proxies for the underlying capacities and transitions.
    Basis for all quantitative diagnostics and indicator transformations in the study.
invented entities (2)
  • Agentic economy no independent evidence
    purpose: To describe the transition toward distributed economic action involving humans, AI agents, robots, protocols, compute, and energy systems.
    Newly developed conceptual framework introduced to organize the analysis.
  • Action-capacity framework no independent evidence
    purpose: To link model/software-agent capacity, robotic capacity, compute-energy coupling, protocolisation, auditable trust, and human sovereignty.
    Introduced as the organizing structure for the contribution.

pith-pipeline@v0.9.0 · 5793 in / 1547 out tokens · 56745 ms · 2026-05-20T01:41:44.940177+00:00 · methodology

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

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

  1. [1]

    All references below include either a DOI link or an official/stable URL. The list is formatted in APA style and is restricted to sources relevant to institutional economics, innovation economics , platform economics, automation and labour, AI systems, robotics, compute-energy infrastructure, AI governance and the empirical source families used by the art...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1016/s0169-7218(11)02410-5 2011

  2. [2]

    On the Opportunities and Risks of Foundation Models

    OECD Publishing. https://doi.org/10.1787/5jlz9h56dvq7-en Arrow, K. J. (1974). The limits of organization. W. W. Norton. https://books.google.com/books?id=8wW4AAAAIAAJ Autor, D. H. (2015). Why are there still so many jobs? The history and future of workplace automation. Journal of Economic Perspectives, 29(3), 3–30. https://doi.org/10.1257/jep.29.3.3 49 Au...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1787/5jlz9h56dvq7-en 1974

  3. [3]

    https://www.koomey.com/post/8323374335 Lasi, H., Fettke, P., Kemper, H

    Analytics Press. https://www.koomey.com/post/8323374335 Lasi, H., Fettke, P., Kemper, H. -G., Feld, T., & Hoffmann, M. (2014). Industry 4.0. Business & Information Systems Engineering, 6, 239–242. https://doi.org/10.1007/s12599-014-0334-4 LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521, 436 –444. https://doi.org/10.1038/nature14539 ...

    work page doi:10.1007/s12599-014-0334-4 2014

  4. [4]

    https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman, L., Hutchinson, B., et al. (2019). Model cards for model reporting. Proceedings of the Conference on Fairness, Accountability, and Transparency, 220 –229. https://doi.org/10.1145/3287560.3287596 National Institute of S...

    work page doi:10.1145/3287560.3287596 2019

  5. [5]

    https://www.weforum.org/publications/the-future- of-jobs-report-2025/

    WEF. https://www.weforum.org/publications/the-future- of-jobs-report-2025/

    work page 2025