Imagine a salesperson at a large software firm preparing to pitch a complex, custom solution to a Fortune 500 client. A decade ago, that preparation might have meant hours of manual research, stitching together case studies, competitive intelligence, and rough drafts of a value proposition. Today, an increasing share of that work can be handed to a generative AI system that produces alternative narratives, use cases, and proposal drafts in minutes. But what happens to the salesperson’s own creativity when the machine is doing so much of the ideation?
That question drives a new study published in the Journal of Personal Selling & Sales Management, which develops a framework for understanding how generative AI reshapes the creative work of business-to-business selling. The central argument: generative AI expands one kind of creativity in sales while leaving another kind squarely in human hands.
The question behind the research
Yuanyuan (Gina) Cui and Patrick van Esch of the E. Craig Wall Sr. College of Business at Coastal Carolina University noticed a gap between how quickly companies are deploying generative AI in sales and how little theory exists to explain what it actually does to selling itself. Sales managers have been reporting uneven results from AI investments, hinting that technical capability alone does not translate into better selling outcomes.
Earlier research on sales technology mostly treated digital tools as automation or decision-support aids, systems that help salespeople work faster or make sharper predictions. Generative AI is different. Rather than retrieving or scoring existing information, it produces new text, imagery, and narratives by recombining patterns from vast training datasets. That means it can participate in the creative parts of selling, framing customer problems, proposing solutions, and articulating value, that older technologies simply couldn’t touch.
The authors wanted to understand how that participation reshapes the salesperson’s role, particularly in complex B2B contexts where deals require customization, extended interaction, and joint problem-solving with buyers.
Two kinds of creativity
To make sense of what generative AI does and doesn’t do, Cui and van Esch borrow a distinction from creativity research: relative creativity versus absolute creativity.
Relative creativity is about recombination. It means generating new variations by remixing existing ideas, narratives, or solution templates. Absolute creativity is different. It involves breaking away from established frames entirely, reframing a customer’s problem in a genuinely new way, or arriving at a solution that doesn’t just recombine what’s already known.
Generative AI, the authors argue, is well-suited to the first kind. It’s built on probabilistic recombination of learned patterns, which is essentially what relative creativity requires. But it lacks contextual judgment, ethical reasoning, and the capacity for intentional problem reframing, all of which absolute creativity demands.
The researchers pair this creativity framework with socio-technical systems theory, which holds that technology creates value only when embedded in human workflows and governed by organizational norms. Together, these lenses form the backbone of the study: socio-technical systems theory explains the conditions under which generative AI produces value in selling, and evolutionary creativity theory explains what specifically it does to the creative work of salespeople.
How the study was conducted
Because generative AI in sales is still a young phenomenon and large-scale sales-specific datasets are scarce, the authors opted for a qualitative, theory-building approach. They analyzed five publicly documented B2B cases involving Salesforce, Adobe, SAP, IBM, and Siemens, firms that all sell complex solutions through relational, consultative processes and have publicly described integrating generative AI into customer-facing sales work.
Data came from company white papers, executive interviews in business publications, product documentation, investor presentations, and practitioner-oriented case materials from consulting firms. The authors cross-checked sources to guard against promotional bias and focused their analysis on selling activities, workflows, and role configurations rather than on companies’ claims about performance gains.
The analysis used what the authors call abductive reasoning, an iterative movement between theory and evidence. Rather than starting with fixed hypotheses to test or building theory purely from the data, they moved back and forth between their theoretical lenses and the case materials, refining their propositions as patterns emerged.
They organized their analysis around three stages of the selling process: preparation (pre-interaction planning), incubation (active buyer-seller interaction), and verification (validating solutions and securing commitment).
What the cross-case analysis revealed
In the preparation stage, a consistent pattern emerged across all five cases: salespeople used generative AI to expand the range of solution narratives and proposal drafts they could bring to a customer conversation. The AI wasn’t producing final answers. Instead, its outputs were treated as raw material that salespeople evaluated and adapted based on what they knew about the customer, the industry, and the competitive situation. In other words, AI broadened the possibilities while human judgment governed selection.
The incubation stage, when salespeople are actively engaging with customers, produced the most variation. In cases where salespeople kept clear interpretive control over AI outputs, drawing on AI-generated framings selectively and adapting them to real-time customer cues, the evidence pointed to enhanced relative creativity without any apparent reduction in deeper creative engagement. But in cases where AI outputs were adopted with little critical evaluation, often under time pressure or in organizations that emphasized speed and output volume, the researchers observed patterns consistent with what they call cognitive offloading: salespeople engaged less deeply with customer-specific nuances and did less real-time problem reframing.
The verification stage revealed something the authors found consistent across all five cases. Customers attributed responsibility for proposed solutions to the human salesperson, not to the AI system, regardless of how much AI had been used in developing the proposal. Salespeople were observed exercising discretion about which AI-generated outputs to use, modify, or reject before presenting them, particularly when ethical alignment, strategic coherence, or relational credibility was on the line. Generative AI could help sharpen a value articulation, but the accountability for securing customer commitment stayed with the human.
The role of managerial governance
Across all three stages, one factor kept surfacing as decisive: how managers set up the rules of engagement. Organizations that had invested in explicit governance, training programs, usage guidelines, and performance frameworks that positioned AI as an augmentation resource, showed patterns of salespeople engaging critically with AI outputs and preserving their own interpretive judgment. Organizations with less deliberate governance, or with norms that prioritized standardization and speed, showed patterns more consistent with reduced salesperson discretion and greater risk of over-reliance on AI.
The authors interpret this as evidence that managerial framing determines whether generative AI functions as a creative partner or as a prescriptive authority in the sales process.
Practical takeaways for sales organizations
The study offers several concrete implications for managers integrating generative AI into their sales operations. The authors caution against treating AI as a substitute for salesperson creativity or judgment. Instead, they suggest designing systems that deliberately support recombination-style creativity while preserving space for deeper, human-only reframing.
In preparation, training should emphasize critical evaluation of AI outputs, treating generated proposals as starting points rather than finished products. During customer interaction, managers should reinforce salesperson discretion over when and how to deploy AI, and establish norms that prioritize understanding the customer over sheer speed. In the verification stage, performance metrics should reward thoughtful value articulation and ethical decision-making rather than raw adoption or frequency of AI use.
Caveats
The authors are transparent about the limits of their work. The study relies on publicly documented cases, which likely overrepresent successful or strategically framed implementations and don’t allow direct observation of salesperson thinking or micro-level interactions with customers. The firms studied are large, technology-intensive B2B organizations, so the findings may not transfer cleanly to transactional selling or smaller enterprises with fewer resources for governance infrastructure.
The propositions the authors develop are just that, propositions, not empirically tested relationships. They call for future quantitative work, including surveys and experiments, to validate whether the patterns they observed hold up under controlled conditions. They also note that customer perceptions of AI involvement in sales interactions, including questions about transparency, authenticity, and trust, remain largely unexplored.
