To LLM, or Not to LLM: How Designers and Developers Navigate LLMs as Tools or Teammates

A grounded theory study of 33 designers and developers reveals that organizational acceptance of LLMs depends on how they're positioned within workflows: as controlled tools versus collaborative teammates. Clear human authority and accountability enable integration, while ambiguous agency creates resistance, suggesting LLM adoption is fundamentally a sociotechnical positioning problem rather than a technical capability question.

This research addresses a critical gap in how organizations approach LLM integration. Rather than evaluating adoption through a purely technical lens focused on model capabilities, the study identifies that practitioners make decisions based on governance and accountability structures. When LLMs function as clearly supervised tools with defined human oversight, teams integrate them readily into existing workflows. However, when roles become ambiguous—treating LLMs as teammates with shared agency—organizational hesitation emerges, driven by concerns about responsibility attribution and liability.

The finding reflects broader patterns in enterprise AI adoption where technical capability rarely determines actual deployment. Organizations have spent decades building hierarchical accountability structures around human decision-making; LLMs introduce ambiguity that conflicts with these frameworks. This is particularly acute in design and development contexts where output quality directly impacts products and user safety.

For the AI industry, this suggests that successful LLM vendors must address governance and compliance concerns alongside technical performance. Tools that provide clear audit trails, explicit human-in-the-loop mechanisms, and transparent decision attribution will see faster enterprise adoption than those focusing purely on capability. For development teams, the research validates that positioning matters—framing LLMs as augmentation tools rather than replacements reduces organizational friction.

Looking forward, this work indicates that enterprise LLM adoption will depend less on model scaling and more on developing governance frameworks that map to existing organizational accountability structures. Companies investing in interpretability, traceability, and clear role definition will outpace those betting solely on raw capability improvements.

• →LLM adoption decisions are primarily sociotechnical positioning problems, not purely technical capability questions
• →Clear human control and accountability enable organizational acceptance; ambiguous agency creates resistance
• →Enterprise LLM success depends on fitting tools into existing governance and responsibility structures
• →Productive teammate configurations remain possible when LLMs operate within explicit oversight mechanisms
• →Tool framing emphasizes control and human authority, while teammate framing introduces problematic accountability ambiguity