Embodied AI Agents for Team Collaboration in Co-located Blue-Collar Work

Blue-collar work is often highly collaborative, embodied, and situated in shared physical environments, yet most research on collaborative AI has focused on white-collar work. This position paper explores how the embodied nature of AI agents can support team collaboration and communication in co-located blue-collar workplaces. From the context of our newly started CAI-BLUE research project, we present two speculative scenarios from industrial and maintenance contexts that illustrate how embodied AI agents can support shared situational awareness and facilitate inclusive communication across experience levels. We outline open questions related to embodied AI agent design around worker inclusion, agency, transformation of blue-collar collaboration practices over time, and forms of acceptable AI embodiments. We argue that embodiment is not just an aesthetic choice but should become a socio-material design strategy of AI systems in blue-collar workplaces.

💡 Research Summary

The paper addresses a gap in collaborative artificial intelligence research by shifting focus from predominantly white‑collar, office‑based settings to the embodied, co‑located environments of blue‑collar work. Drawing on the newly launched CAI‑BLUE project (2026‑2029), the authors argue that AI agents should not be treated merely as productivity tools for individuals but as visible, physical teammates that can enhance shared situational awareness, inclusive communication, and on‑the‑job learning. Two speculative scenarios illustrate potential applications. In a smart factory, an embodied AI agent monitors sensor streams, diagnoses a mechanical fault, and projects its analysis to the entire assembly team via visual and auditory cues. Mid‑career worker Elina and her colleagues discuss the suggestions, decide on a corrective action, and the AI updates its model based on the chosen solution, thereby reinforcing collective cognition and reducing downtime. In a car‑repair shop, an embodied transcription system listens to a conversation between a veteran mechanic and a novice, automatically fills out service reports, and later serves as a searchable knowledge base for regulatory and procedural queries. The physical presence of the AI—whether a robot body, a tangible display, or a spatial speaker—allows it to participate in the workflow without diverting attention to personal devices, preserving face‑to‑face interaction while providing hands‑free, conversational support.
Beyond benefits, the authors outline four open design questions. First, how to prevent embodied agents from becoming surveillance tools; this requires explicit worker control over data collection, transparent usage policies, and clear boundaries on what is recorded. Second, the distribution of agency and authority among workers, management, and the AI; embodiment choices (voice, placement, embodiment style) can signal hierarchy and must be calibrated to maintain worker autonomy. Third, the longitudinal impact on collaboration practices; longitudinal field studies are needed to track how workers adapt, how new tensions emerge, and how the AI reshapes social dynamics over time. Fourth, cultural and contextual acceptability of different embodiments; safety regulations, cultural norms, and power relations differ across factories, maintenance sites, and care environments, demanding participatory design to align AI form factors with local values.
The conclusion reiterates that embodiment is not a cosmetic add‑on but a socio‑material design strategy essential for supporting genuine human‑human collaboration in blue‑collar contexts. Responsible, context‑sensitive design—grounded in participatory methods, in‑the‑wild deployments, and mixed‑method evaluations—will be crucial to ensure that embodied AI agents enhance shared understanding, inclusivity, and coordination while respecting workers’ dignity, expertise, and agency. The CAI‑BLUE project will operationalize this agenda through iterative research‑through‑design cycles, field trials, and rigorous evaluation of lived worker experiences with embodied AI.