How do managers in small craft businesses delegate work to generative AI?

Host: Welcome to A.I.S. Insights — powered by Living Knowledge. I'm Anna Ivy Summers.

Expert: And I'm Alex Ian Sutherland.

Host: Today we're exploring a fascinating study titled "Manager-AI Agency Dyads in Skilled Crafts: Configuring Delegation Patterns." Now, Alex, when most people think of Generative AI, they picture tech startups or massive corporations. But this study investigates blue-collar, craft-based small and medium-sized enterprises. What is the core focus here?

Expert: It really challenges conventional thinking, Anna. The study examines how managers in craft businesses—like carpentry, electrical work, and metal fabrication—delegate tasks to Generative AI. By looking through the lens of context-dependent delegation theory, the authors analyzed how these managers form agency dyads with AI to distribute implementation, cognitive, and communicative functions.

Host: That brings us directly to the real-world problem this study addresses. In mainstream tech discussions, people often assume AI autonomy grows in a straight line, eventually taking over entire managerial roles. Why is that assumption flawed, especially for craft businesses?

Expert: That linear view simply doesn't hold up in practice. In craft-based SMEs, managers operate under intense resource constraints. They are typically balancing hands-on manual work on the shop floor or job site with complex administrative duties, often without formal business or management training. They can't just hand over complete autonomy to an AI system. The risk to their reputation, craft standards, and customer relationships is far too high. Prior research didn't explain how delegation actually works when managers need AI assistance but cannot afford to lose control.

Host: They need support, but they can't abdicate responsibility. So how did the researchers investigate how these managers make it work?

Expert: They conducted a rich qualitative study, analyzing in-depth interviews across sixteen craft-based SMEs in Germany. They looked at managers who actively use Generative AI tools like ChatGPT in their day-to-day operations. Rather than treating delegation as an all-or-nothing decision, they tracked how specific functional components—thinking, drafting, and communicating—are divided between the human manager and the AI.

Host: And what did they find? How are these craft managers actually setting up their collaboration with AI?

Expert: The study revealed three distinct configurations of the Manager-AI dyad: Execution, Sparring, and Representation.

Host: Let's walk through those. What does the Execution pattern look like?

Expert: In the Execution configuration, the manager retains full cognitive control. They define the intent, structure, and parameters of a task, and use AI purely as an efficient drafting engine. For example, a manager might outline the exact technical scope of a project and ask the AI to generate a structured proposal. The AI handles the heavy lifting of writing, but the human directs the content and verifies the output.

Host: That sounds like a powerful time-saver. What about the Sparring configuration?

Expert: Sparring is deeply collaborative and cognitive. Here, the manager uses the AI as a conversational thinking partner or sounding board. If a manager needs to structure an uncomfortable employee review, brainstorm a new service offering, or think through a complex pricing model, they engage in a dialogue with the AI to refine their own logic and uncover blind spots. It's essentially an on-demand advisor.

Host: That is remarkably strategic for managers who typically operate in isolation. And what is the third configuration, Representation?

Expert: In Representation, the AI acts as a bounded communicative proxy toward external stakeholders. Many craft managers excel at technical execution but struggle with drafting polished, diplomatic customer emails or public marketing copy. With Representation, the AI crafts the external voice, translating technical notes into client-friendly language. However—and this is a critical finding of the study—the manager always acts as the final gatekeeper, ratifying the message before it ever reaches an external party.

Host: That highlights one of the study's biggest conclusions: working with AI does not mean handing over decision-making authority.

Expert: Exactly. AI integration in these firms is not a wholesale transfer of authority. It is a selective allocation of specific functional tasks. Human managers consistently retain ultimate decision-making, oversight, and accountability.

Host: Let's pivot to why this matters so much for business professionals and leaders listening today. What are the key takeaways?

Expert: First, it completely demystifies AI adoption. You don't need an enterprise-scale transformation or autonomous agents running wild to see massive value. For resource-strapped leaders, tailoring AI delegation allows them to dramatically reduce cognitive overload and administrative burden.

Host: It essentially allows leaders to expand what is feasible for a small team to accomplish.

Expert: Precisely. A small business can now produce professional marketing, refine internal strategy, and communicate at a scale previously reserved for large firms with dedicated departments. Most importantly, it shows that keeping the human in the loop isn't an operational bottleneck; it is the deliberate design choice that protects quality, trust, and authenticity.

Host: A brilliant reminder that AI should amplify human expertise, not replace it. To summarize, this study shows that by strategically configuring AI into roles of Execution, Sparring, and Representation, managers can scale their capabilities while maintaining firm control over their core business. Alex, thank you for sharing your analysis today.

Expert: Thank you, Anna.

Host: And thank you to our listeners for tuning in to A.I.S. Insights — powered by Living Knowledge. Join us next time as we unpack the research shaping the future of business and technology.