Read the original article in Harvard Business Review
Why strategic thinking has been constrained
Csaszar starts with bounded rationality: even capable leadership teams have limited time, attention and capacity to compare alternatives. Familiar strategy frameworks help people simplify complex decisions, but simplification also leaves possibilities unexplored.
He identifies three ways AI can widen strategic thinking:
| Strategic task | What AI makes possible | A B2B customer question worth exploring |
|---|---|---|
| Search for alternatives | Consider more potential directions before narrowing the choice. | Which customer problems, roles or markets have we not examined? |
| Represent the environment | Build richer models of customers, companies and market conditions. | What would make this change relevant in a particular customer’s business? |
| Aggregate and challenge judgments | Use synthetic deliberation and structured critique to test assumptions. | Which stakeholder might question the value, feasibility or timing of our proposal? |
The first two activities broaden what a team can see. The third helps it challenge what it thinks it knows. Virtual Customer is relevant especially where those activities depend on understanding the customer side of a B2B decision.
The customer perspective needs a business world
A generic AI persona may sound convincing while knowing little about the conditions under which a decision is made. A job title alone does not explain the responsibilities, dependencies or competing priorities of a customer organization.
Virtual Customer begins with a business world model: a defined context spanning the customer company, its market and the wider environment. The team specifies four central dimensions: industry, region, company size and role.
Within that world, a virtual customer’s perspective is shaped by its responsibilities, goals, knowledge and constraints. An operations leader considering a maintenance service may focus on changes to daily work. A CFO may examine financial priorities. Procurement may consider supplier requirements. None should be treated as an all-knowing spokesperson for the entire company.
Teams can explore these perspectives through text or voice conversations, compare what each role brings into question and return to the same virtual customers as the offering changes. For help selecting relevant roles, see B2B stakeholder mapping. Csaszar also describes living models that evolve as conditions change. Virtual Customer supplies contextualized customer perspectives, but persistent conversations do not imply an automatically refreshed or perfectly current model of a particular company.
A practical customer-perspective challenge
Before a consequential B2B product or market decision, try this lightweight exercise. It adapts Csaszar’s three strategic tasks into a customer-learning workflow.
- State the decision, not just the idea. For example: should an established industrial software product add a predictive maintenance workflow or expand its current reporting service?
- Keep the context comparable. Define the target industry, region and company size. Hold those dimensions and the customer problem steady while exploring different roles or alternatives.
- Start with current work. Ask virtual customers what responsibilities, existing alternatives and pressures could shape this decision before presenting either proposal.
- Introduce and challenge the alternatives. Explore potential benefits, trade-offs and objections separately with the roles that would use, approve, fund or implement the change.
- Record what needs evidence. Note competing explanations and the questions that could change your decision. Then select customer interviews, observation or a pilot suited to those uncertainties.
The output is a set of better questions and clearer decision assumptions, compiled and interpreted by your team, not an automated verdict on which strategy will win.
Example: two directions for an established industrial service
A B2B software company already provides equipment-monitoring reports. It is considering either a more interactive maintenance-planning workflow or an AI-assisted recommendation service for midsize manufacturing companies with multiple sites in Germany.
The product team creates a consistent customer business context and explores relevant virtual roles: an operations leader responsible for maintenance work, a CFO concerned with investment priorities, a CTO involved in technical adoption, and a procurement lead concerned with supplier processes.
Rather than asking every role, “Which option do you prefer?”, the team examines the work and decision that each alternative would change. The team compares the concerns raised across roles, keeping simulated responses separate from customer observations.
The team revises the proposal after the first exploration. It returns to those virtual customers with the changed workflow, preserving the earlier discussion as context. The next research step is customer interviews about current maintenance decisions or a pilot to observe whether the proposed workflow helps.
Synthetic deliberation should challenge a decision, not certify it
Csaszar describes creator, critic and competitor agents, along with simulated stakeholders, as ways to introduce structured challenge into strategy. Virtual Customer contributes a specific customer-facing part of this approach: examining how a defined B2B role might interpret a problem, proposal or change.
That is different from claiming the platform automatically runs a complete strategy war game, assembles a representative synthetic panel, models competitor behavior or predicts what customers will buy. A modeled objection is a reason to investigate, not customer evidence in itself.
This distinction makes the process more useful. A virtual customer can suggest that a team has overlooked an implementation responsibility or misunderstood who owns a problem. The team must then determine what is plausible, what is supported and what needs to be checked with customers. A customer evidence map can help separate those sources.
Build a better strategic decision process, one customer question at a time
Csaszar also argues that competitive advantage comes less from access to the same AI models and more from the distinctive data, processes and capabilities organizations build around them.
Csaszar calls for a hybrid strategist who frames questions and knows where AI-supported analysis must yield to human judgment. For a B2B product team, one such capability is the habit of returning to customer questions throughout the product and service lifecycle. Virtual Customer offers a repeatable way to explore contextualized, simulated stakeholder perspectives between and alongside customer research.
Start with one decision. Choose the industry, region, company size and two customer roles that matter. Explore the current problem before introducing your proposed change. Compare the perspectives, then identify one assumption whose resolution could alter your next investment or product decision.
Explore your next strategic customer question with Virtual Customer.