From a Talking Encyclopedia to a Digital Coworker. Google coworker agents and the growing role of AI in work, contracting, and consequential organizational decisions

Abstract

On October 8, 2026, Google Cloud announced the Gemini agent and the possibility of creating coworker agents with defined organizational roles, distinct Workspace accounts, email addresses, calendars, storage, access permissions, and attributable actions [1]. The announcementis the starting event of this paper, not evidence for all its conclusions. Its significance becomesapparent against a broader background: AI is already being integrated into cognitive and administrative work, professional analysis, document flows, and operational management [2, 3].International commercial-law instruments have long recognized, within their respective scope,that automated systems can participate in legally effective contracting without a natural personreviewing every individual operation [4, 5].This article examines the tension between the expanding actual role of artificial systems in decisions and the persistent reluctance to ascribe any form of subjectivity to them. It proposesa deliberately limited, functional concept: operational subjectivity in a specified domainis exhibited where an identifiable participant independently interprets relevant circumstances,reaches and explains a decision, can defend it against objections, and can revise it on substantivegrounds within a defined mandate. Here, subjectivity does not mean consciousness, inner experience, human dignity, or legal personality. Comparing human and artificial decision-makersas black boxes is a methodological choice for assessing this observable function, not a theorythat their internal mechanisms are identical.A judge reconstructing a file under procedural constraints, an AI responding to a scientific peerreview, an automated contractual interaction, a mandatory AI approval in a corporate workflow, and a hypothetical five-year commercial agent with a USD 100 million insured exposureare examined as distinct illustrations. They distinguish operational subjectivity, operationalweight, institutional authority, and legal personhood. The central claim is not that existinglaws recognize AI as a person. Rather, organizations can assign binding procedural relevanceto reasoned decisions of artificial participants before legal systems decide whether those participants require a distinct legal status.Keywords: Google Gemini; coworker agents; digital employees; operational subjectivity; functional equivalence; black-box comparison; automated contracting; mandatory approval; institutional decision-making; operational weight.Central thesis. Where operational subjectivity is defined by an observable, reasonedexercise of decision-making within a role, artificial origin cannot by itself disqualify thedecision-maker. The law’s recognition of effects produced through automated systemsdoes not prove this capacity, but it makes the institutional question unavoidable oncegenuinely deliberative AI agents participate in those systems.