By Mostafa Sayyadi and Luca Collina
As AI reshapes advisory work, the real value of management consultants is shifting from analysis to interpretation, relationships, judgment and accountability.
From financial services to management consulting, artificial intelligence is transforming advisory work. While existing research focuses on adoption, productivity and human–AI collaboration, it offers limited insight into how advisory value is being reconfigured. The “advisory stack” provides a layered framework in which AI and professionals share analytical generation, interpretation and relational engagement, while human judgment and accountability remain central to decision-making in an AI-enabled advisory environment.
Advisory work has long been built on a relatively stable model. Clients want; experts consult — structured analysis and recommendations turn into solutions that enhance value through insight and interpretation in a climate of trust. Analytical capability and structured reasoning are rare, thus valuable for this model. Artificial intelligence shatters this assumption by democratizing analytical capability at scale. AI systems today write analyses, organize problems, and develop recommendations at a scale and speed previously impossible for humans. For example, robot-advisors automate decisions regarding portfolio allocation and risk management in the financial services domain. Career impacts. In management consulting, generative AI tools facilitate research and synthesis and the articulation of findings to clients — particularly in the early stages of analysis for hypothesis generation. Thus, the line between human and AI advice becomes a benediction.
The crux of the debate has centered on whether AI-based solutions will replace human advisors. This framing is insufficient. Advisory work itself is not going away, but is undergoing a structural reshuffling. One of the biggest changes is not substitution but rather redistributing what we do and how we do it with much more value. Hybrid configurations, in which AI systems and consultants coexist within the same decision-making process with a client (Jung et al., 2018), are emerging. This paper presents the advisory stack, a socio-technical decision framework that extends traditional advice and decision-making theories by modelling how value and responsibility move vertically across layers. To clarify: This framework introduces no new elements; it merely orders existing concepts into a structure that explains what advisory work looks like in the age of AI. It argues that the profession of managing consultants’ responsibility shifts from creating solutions to governing how decisions are modelled, justified, and validated.
Foundations of Advisory Work
Traditionally, advisory has been understood as expertise + interpretation + relational influence. The advisor helps clients find their own understanding and solutions to their problems rather than simply providing them. The real value of a professional service firm comes not only from technical competence but also from trusting relationships and longevity. Professional organizations intertwine knowledge-based capabilities with institutional and relational mechanisms. These perspectives establish advisory work as necessarily multi-material. But they suppose that all dimensions are created for humans by human professionals. The underlying assumption becomes unstable when analytical and informational tasks can be performed by AI systems.
Human vs AI Advisory Debate
In the past, advisory work was conceived as a mix of expertise, interpretation, and relational power. A process consultation approach focuses on the problem’s identity and helps clients understand their own situation, rather than simply dispensing solutions. Professional service firms create value through professionalism, technical skills, trust, and the ability to form long-term relationships. Professional organizations integrate knowledge-based capabilities with institutional and relational mechanisms. These perspectives establish advisory work as multi-dimensional by its nature. But they assume that all dimensions are delivered by people. When AI systems can perform analytical and informational tasks, this assumption is no longer valid.
Emerging Hybrid Advisory Models
There is also well-documented prior research on AI in advisory work, but such evidence does not explain why and how advisors’ functions are reorganized as a system. There is no cross-practice framework. The literature prioritizes efficiency while neglecting judgement and responsibility. In the 1960s, our understanding of roles was still very much one-dimensional rather than multi-layered, and what it meant to be a managing consultant evolved slowly (and remains poorly defined).
The Advisory Stack: A Conceptual Framework
The advisory stack considers this advice work to be a system of layers with different sets of values and obligations tied to how they interrelate. The bottom of the analytical generation process consists of information processing, modelling, and option creation. This layer is continually being automated by AI systems capable of scaling and operating at unprecedented speeds. This level often produces voluminous raw discharges quickly, but these discharges require further processing before they can be applied in organizational settings.
Interpretation places analytical outputs in context-specific knowledge; it is layered on top of this. This is an expression of within-task complementarity, whereby the combination of human judgement with AI output yields outcomes that are better than either acting alone. This layer is where the managing consultant comes into play, sifting through information flows, ranking and restating outputs relative to strategic goals and limits.
Relational engagement links analysis with organizational dynamics, stakeholder expectations and the articulation of more effective ways of communication. This layer mirrors the limitations of automation that some scholars point to in tacit knowledge, as well as Section 4 on AI performance being a “jagged frontier” where nuanced, context-dependent tasks are not where systems excel. Especially in complex or politically sensitive environments, trust, persuasion, and alignment remain critical. Performance. Judgment and accountability validate decisions with the role of assigning who is responsible at the top. This layer corresponds with the Judge–Advisor System, whereby a human expert is considered to have the final say over whether or not to follow machine-made recommendations.
The interaction between these layers is continuous (as opposed to sequential). Layer-to-layer interactions are regularly dynamic and adaptive. Automation leads to more data and faster processes, but output has little value without a clear contextual understanding of what it signifies. Too much analytical power without enough interpretative capacity can overwhelm your decision-making, rather than match it. On the other hand, powerful interpretative powers can offset deficits in analytical might by shining a light on the most relevant signals. These dynamics are carried through in terms of relational engagement. Analytical insights only impact decisions if they are expressed in whatever form meets stakeholder expectations and is also aligned with organizational priorities.
This means that the layers have to keep adjusting all at once. AI outputs inform interpretation, and the way we engage informs how decisions are validated. Feedback loops are a common occurrence as decisions produce more data to analyses. The power of the advisory stack resides less in which individual layers are strong but rather in how well they march together. Say an AI system produces strategic choices through macro data analysis, but the overarching role of a managing consultant is to screen those against organizational limits and stakeholder interests or risk exposure before settling on any decision. AI-generated alternatives will be dependent on the boundaries of your organization and where stakeholder alignment lies in any strategy engagement. This is where accountability sits.
Layer misalignment may cause entry mistakes wherever the output from an automated technique isn’t reviewed or evaluated. One of the ways that things fail is for one layer to rule them all. A typical failure is considering an analytical output as ready, leading to decisions that might seem data-led but are not contextualized. Another happens when interpretation becomes untethered from the underlying data, leading to conclusions that sound convincing but are not evidence-based. Outcomes can additionally be threatened through relational failures. Resistance can bubble up when stakeholder alignment is assumed instead of managed at the point of implementation. A more insidious risk can be the diffusion of accountability. Responsibility is ambiguous; decisions may be passed on to technical systems or diffused across a constellation of actors. This is why, at the level of judgment, ownership must be absolutely crystal clear.
Figure 1. The Advisory Stack: How value and responsibility are distributed across layers.
The graphic depicts advisory work as a systematized stack: AI underpins analytical generation, managing consultants – led by the determination of appropriate interpretation and relational engagement; accountable judgment is co-shared between executive leadership and clients, where responsibility ultimately converges. The layered structure rests on established theory: analytical generation for AI as decision support, interpretation for complementarity within tasks, and relational engagement because of limitations in tacit knowledge and judgment following the Judge–Advisor System.
Implications for Managing Consultants
The advisory stack redefines the role of management consultants in seeing their new functions. They govern the conditions of decision-making rather than producing analysis. This means structuring problems, tying out layer outputs, and connecting analytical insights to an organization’s context. Such a shift necessitates a change of capabilities. Managing consultants will need the ability to comprehend AI-generated outputs, their implications, and limitations. Moreover, they need to cope with expectations from stakeholders about the decisions being robust and legitimate.
It defines where value is created in advisory work. With AI cranking down the difficulty of analytical tasks, both time and effort become far less reliable proxies for value. Focus shifts up the Value Chain toward interpretation, relational engagement, and judgment. This means less valuing billable hours as a proxy for value, and more focus on advisory work against decision quality (and, to the extent relevant, downstream observable outcomes), requiring managing consultants who structure decision processes, validate outputs, and hold stakeholder parties accountable.
Discussion
The advisory stack redefines the role of management consultants by introducing new functions. They govern the conditions of decision-making rather than producing analysis. This means structuring problems, tying layer outputs, and connecting analytical insights to an organization’s context. Such a shift necessitates a change of capabilities. Managing consultants will need the ability to comprehend AI-generated outputs, their implications, and limitations. Moreover, they need to cope with expectations from stakeholders about the decisions being robust and legitimate.
It gives a definition of where value is created in advisory work. With AI cranking down the difficulty of analytical tasks, both time and effort become far less reliable proxies for value. Focus shifts up the Value Chain toward interpretation, relational engagement, and judgment. This means less valuing billable hours as a proxy for value, and more focus on advisory work against decision quality (and, to the extent relevant, downstream observable outcomes), requiring managing consultants who structure decision processes, validate outputs, and hold stakeholder parties accountable.
In Conclusion
AI does not eliminate advisory work; it redistributes it. The advisory stack provides a framework for understanding how value and responsibility are reorganized in AI-enabled environments. Future research could test how advisory value is distributed across layers in different industries and organizational contexts, providing empirical validation of the framework.
Mostafa Sayyadi
