Beyond the Pyramid: AI and the New Architecture of Consulting Firms
In the first article of this series, When Algorithms Advise: AI's Disruption of the Consulting Model, we set out how artificial intelligence is changing the advisory value chain. Generative tools now handle research and synthesis at volume, agents complete connected tasks, and clients are building analytical capability inside their own organisations.
We also described the model that change is testing. The consulting pyramid, analysts at the base and partners at the apex, was built on two conditions: that human hours were the only way to scale analysis, and that firms held information clients could not easily reach themselves. AI weakens both.
This second article asks what the firm looks like when those conditions no longer hold. Hierarchy does not disappear. It still allocates responsibility, develops people and signs off on advice. What it loses is its monopoly, because knowledge, work and accountability now move through systems as well as through tiers.
That shift plays out across four areas: how consulting is priced, how trust is established when part of the analysis is machine generated, how firms organise roles and governance around systems as well as people, and which principles should guide the redesign.
One relationship sits underneath all four:
Intelligent Reinvention = AI Capability × Human Intent × Measurable Impact
AI capability provides scale, speed and analytical reach. Human intent defines the problem, applies judgement and accepts responsibility. Measurable impact shows whether the work created value. The multiplication is the point. Strong technology produces little when the purpose is unclear, good judgement does not scale through unreliable systems, and neither counts commercially if the result cannot be demonstrated.
For consulting firms, the question is no longer whether to use AI. It is how to rebuild around it without losing the judgement clients are paying for.
From "Sell Time" to "Sell Outcomes": The Economic Reinvention
For more than a century, consulting has rested on a simple formula: revenue equals time multiplied by expertise. Firms scaled by leveraging human hours, building pyramids of analysts and managers to gather information, run analysis and produce recommendations. The partner's judgement was the premium layer, supported by an organised process of research and synthesis.
AI puts pressure on that equation. When parts of the analytical process can be completed faster, hours become a weaker proxy for the value created. Clients still pay for expertise and delivery capacity, but they increasingly evaluate engagements through measurable business impact, implementation progress and the capabilities left behind.
The Billable Hour Under Pressure
Generative and predictive AI compress significant parts of the delivery cycle. Scenario analysis, customer segmentation, process mapping and competitive benchmarking can now be supported by automated systems, reducing the manual effort behind some forms of research and synthesis.
Boards and procurement teams have started asking the obvious question: if the model can do this in hours, why am I paying for weeks?
The historic economic lever of consulting, human effort multiplied through hierarchy, is being supplemented by platform leverage. Proprietary data, reusable methodologies, AI systems and digital assets allow firms to apply expertise more consistently across engagements. This is not a cost reduction exercise. It changes how consulting value is created, priced and demonstrated.
Clients Are Recalibrating Their Expectations
Traditional deliverables remain useful, but clients now expect them alongside clearer evidence, faster feedback and closer links to implementation. They ask how findings were produced, how recommendations change as new information arrives, and how progress will be measured once the initial analysis is complete.
The answer to what they are paying for is not speed. It is the quality of the decision, the relevance of the recommendation, the firm's ability to execute and the improvement that follows. Consultants are therefore under pressure to move beyond presenting insight and towards embedding knowledge, tools and processes inside the client's operations.
The Shift Towards Hybrid and Outcome-Based Models
Professional services firms are reconsidering pricing built mainly around time and materials. The direction is not one universal replacement, but a wider combination of structures:
Fixed-fee models, where scope and deliverables are agreed in advance.
Subscription or continuous advisory models, providing ongoing access to expertise, monitoring tools or AI-supported analysis.
Performance-linked arrangements, where part of the fee depends on defined indicators such as cost savings, customer satisfaction or operational improvement.
Managed-service models, where the consultancy becomes embedded in an ongoing business process.
AI makes several of these easier to deliver, because it supports continuous monitoring, repeatable analysis and more frequent reporting.
The limit is control. Results depend on market conditions, internal execution and decisions the consulting firm does not own, which is why hybrid structures are the more immediate change rather than a wholesale move to outcome-based fees. Firms can combine fixed fees, subscriptions, time-based elements and performance incentives according to the responsibility they actually hold.
A New Economic Contract Between Firms and Clients
These models create a more transparent relationship. Clients gain visibility into how work is produced, how progress is assessed and which outcomes the firm is prepared to stand behind. Firms move closer to the client's operational environment, participating in the creation of value rather than advising from the sidelines.
Consulting becomes less dependent on the volume of human effort and more connected to the reliability of the firm's systems, the relevance of its expertise and its ability to help clients act.
Human Oversight and Ethical Guardrails: Redefining Trust
As consulting firms shift from selling time towards delivering outcomes, trust, the foundation of the advisory relationship, faces a new test.
In the traditional model, trust rested on the credentials, experience and judgement of the consultant. Clients did not see every analytical step, but they trusted the people and institutions behind the recommendation. In an AI-supported model, algorithms participate in research, analysis, forecasting and recommendation development. That raises a harder question: how is trust preserved when part of the advisory process is carried out by a system?
From Human Reputation to System Reliability
When clients ask how a recommendation was produced, the answer now includes the data used, the model selected, the instructions given to it and the human review applied to its output.
AI systems process information quickly and identify patterns across large datasets, but they carry no professional accountability. They produce inaccurate, incomplete or contextually inappropriate outputs, particularly when the underlying data, assumptions or instructions are weak. A recommendation can look analytically convincing while missing commercial realities, organisational sensitivities or reputational consequences.
Human credibility therefore remains essential, but it now has to be supported by reliable systems and clearer evidence of how conclusions were reached. Firms need traceable analytical processes, documented assumptions and honest communication about the limits of the systems they use. Not every model can reveal a complete reasoning path, but firms can still be transparent about inputs, methods, validation procedures and the role of human review.
The deliverable does not have to become an interactive system. It does have to be supported by audit trails, data sources, scenario assumptions and monitoring tools that show clients how findings were developed and how they may change.
The Rise of Ethical Guardrails
AI in consulting creates opportunity and liability in the same movement. Incorrect or biased outputs contribute to financial losses, discriminatory decisions, privacy breaches and reputational damage.
The governance challenge grows as systems move beyond generating content and begin recommending or initiating actions. McKinsey's research on AI trust found that only around 30% of organisations had reached a relatively advanced level of maturity across strategy, governance and agentic AI controls, and that organisations with clearly assigned responsibility for responsible AI scored higher than those without explicit ownership.
Responsible AI therefore has to be translated into operating procedure rather than stated as intent. The controls that matter:
Transparency and explainability: Clients should know when AI has materially contributed to an analysis, which information shaped the output and which limitations remain.
Accountability and attribution: Responsibility for AI-supported work should be assigned. Firms must define who validates outputs, who approves their use and who answers when a system produces an unreliable result.
Bias detection and mitigation: Data, assumptions and outputs should be reviewed for patterns that unfairly disadvantage particular individuals, groups or markets.
Data provenance and protection: Firms should maintain records of data sources, permissions and processing decisions, particularly where client, personal or commercially sensitive information is involved.
Ongoing monitoring: Controls should continue after deployment. Models, data and business conditions change, so systems need review for declining accuracy, unexpected behaviour and emerging risks.
Quality assurance expands accordingly. Reviewing presentations and calculations remains necessary, but firms must also examine the systems, data and controls behind the work.
The Regulatory Horizon
The regulatory environment is formalising some of these expectations.
The European Union's AI Act follows a risk-based approach. It does not automatically classify every advisory or decision-support system as high risk. Classification depends on the system's intended purpose and whether it is used in specified areas that may seriously affect health, safety or fundamental rights, such as employment, access to essential services, law enforcement or the administration of justice.
The timeline has moved. General-purpose AI obligations have applied since August 2025, and the Act's transparency rules are taking effect through 2026. Under the Digital Omnibus amendments, the stricter requirements for high-risk systems, covering documentation, activity logging, risk management and human oversight, now apply from 2 December 2027 for stand-alone systems and 2 August 2028 for AI embedded in regulated products.
What this means in practice depends on how a system is used. A general research assistant and a system influencing employment, credit or another sensitive decision carry very different obligations. Firms need to assess use cases individually rather than treating all AI-supported advice as one regulatory category.
Compliance is part of the trust architecture, not the whole of it. A firm can meet every legal minimum and still fail to explain its systems, prepare for incidents or earn client confidence. In an environment where AI amplifies both insight and error, trust becomes a product feature.
The Human in the Loop: From Oversight to Orchestration
The answer is not to remove humans from the process. It is to define their role more precisely.
Consultants will keep performing analysis, but more of their value comes from deciding where AI should be used, testing the quality of its outputs and connecting analytical results to commercial and organisational reality.
Oversight should be proportionate to the decision. A low-risk internal research task requires a different review process from a recommendation affecting employment, investment, pricing or market access. What matters is that human involvement represents real judgement rather than a signature added to an unexamined process.
Consultants increasingly act as orchestrators between data, systems, specialists and decision-makers. AI provides speed, scale and pattern recognition. Humans provide intent, accountability and an understanding of consequences that no measure of model performance captures.
The firms that benefit most from AI will not be those that deploy it fastest. They will be those that build clear accountability, reliable controls and enough transparency for clients to understand when, where and why AI shaped the advice they receive.
New Organisational Roles and Governance Structures: The Post-Pyramid Firm
The post-pyramid consulting firm is a network of intelligence systems orchestrated by humans. Analysts, managers and partners still hold important roles, but research, analysis and synthesis are now distributed across consultants, AI tools, specialist systems and shared data platforms.
AI does not remove hierarchy. It changes how work is allocated, how teams are formed and how responsibility is managed, producing a more flexible structure rather than the immediate collapse of the pyramid.
From Hierarchies to Hybrid Architectures
Consulting firms are combining traditional teams with smaller, cross-functional groups of strategists, industry specialists, data scientists, AI engineers and risk professionals. These groups are assembled around the needs of an engagement instead of a fixed staffing pyramid, with AI processing information, automating parts of the workflow and making knowledge available across every level of the firm.
Agentic AI extends this model by allowing systems to plan and complete several connected tasks. Adoption remains early. McKinsey found that 23% of surveyed organisations were scaling an agentic AI system in at least one function, while another 39% were experimenting with the technology. Agents are an emerging capability, not a replacement for an operating model.
New Responsibilities in an AI-Enabled Firm
The post-pyramid firm requires responsibilities that connect technology, delivery and governance. Larger firms turn these into dedicated roles. Smaller organisations attach them to existing positions.
AI product ownership: Managing the development, testing and improvement of internal or client-facing AI tools.
AI integration: Connecting models, agents, data sources and business systems so they work together reliably.
Model risk management: Monitoring accuracy, bias, performance changes and version control.
Ethics and compliance: Turning regulatory and ethical requirements into practical controls.
Outcome design: Connecting project activities to clear operational and commercial measures.
The change underneath all five is the same. Firms now manage not only people and projects, but the systems that contribute to their work.
Governance in an AI-Enabled Firm
As AI takes on more of the delivery, governance has to be built into the operating model.
Traditional approval processes were not designed for systems that update frequently or operate across several workflows. Firms need clear decision rights that define:
who approves an AI use case
who owns the system
who validates its outputs
who monitors its performance
who is accountable when problems occur
The structure carrying those decisions should reflect the size of the organisation, the risks involved and the importance of the decisions being supported. Deloitte's operating-model research argues that organisations scaling AI need to reduce fragmentation, clarify decision rights and improve coordination between business, technology and risk functions.
Dashboards and audit trails support this work without replacing human review. Their purpose is to make system performance, material decisions and identified risks easier to trace.
From Managing People to Coordinating Intelligence
Leadership in consulting still involves managing people. It now also requires coordinating different forms of expertise.
Partners and managers decide which work belongs with consultants, which tasks AI can support and where specialist review is required. They also have to ensure that outputs from different people and systems converge on one clear and accountable recommendation.
The capabilities this rewards are human ones. PwC's 2026 Global AI Jobs Barometer found that AI-exposed roles are adding tasks that rely on empathy, judgement and creativity faster than less-exposed roles. As systems absorb more of the analysis, the work that remains concentrates in the areas where people are hardest to substitute.
The post-pyramid firm is not a firm without hierarchy. It is a firm where hierarchy is no longer the only way that knowledge, work and responsibility are organised.
The Six Principles for Reinvention: Consulting's Blueprint for the AI Age
The pressure on the traditional consulting pyramid does not signal the end of consulting. It shows that firms must reconsider how they create value, organise work and support clients.
This requires more than adding AI tools to existing processes. Deloitte's State of AI in the Enterprise 2026 found that 34% of surveyed organisations were using AI to transform products, services, processes or business models, while 30% were redesigning important processes and 37% were still using AI with little change to how work was organised. The difference lies between improving the existing model and redesigning it.
Adapted from Forbes' Six Principles for Reinvention: Management Consulting in the Age of AI, the following framework sets out what that redesign involves.
1. Co-Creation: Building With Clients, Not for Them
AI-supported consulting should not become a one-way transfer of automated recommendations.
Client teams hold knowledge about their customers, operations, constraints and internal decision-making that external consultants cannot fully reproduce. Co-creation brings that knowledge into the design, testing and improvement of a solution.
Instead of presenting a completed answer at the end of an engagement, consultants and client teams test assumptions together, examine different scenarios and adjust recommendations as new information appears. This gives clients a clearer understanding of how the solution works and increases the likelihood that it will be adopted. The consultant still provides independent judgement, but the client shapes how that judgement is applied.
2. Collaboration: Connecting Business, Technology and Risk
AI-enabled consulting requires expertise from several disciplines.
A market expansion project may need commercial strategy, customer research, data analysis, technology integration, regulatory knowledge and operational planning. These areas cannot run as separate tracks if they are contributing to the same recommendation.
Deloitte's operating-model research argues that scaling AI requires clearer coordination between business, technology, data and risk leaders. Without shared decision rights and accountability, specialist knowledge stays fragmented even when the technology itself is connected.
Collaboration therefore means more than adding technical specialists to a traditional project team. It requires a common problem definition, shared measures of success and a clear process for resolving conflicting evidence or priorities. AI helps teams access and compare information, but it cannot decide which commercial, technical or ethical consideration should carry the most weight. That remains a human responsibility.
3. Modularity: Building Services That Can Be Reused and Adapted
Consulting engagements do not all need to be delivered as large, fixed programmes.
Some client needs are addressed through smaller components that work together: market-monitoring tools, scenario models, customer-segmentation systems, risk assessments or decision dashboards. A modular approach lets firms reuse proven methods while adapting them to the client's market, data and operating conditions. Clients begin with a defined problem, test the value of the solution and expand it when the evidence supports further investment.
Larger programmes remain necessary when several functions, markets or systems must change together. The difference is that the engagement is designed as connected stages rather than one fixed package, with each component producing a useful result while contributing to the wider transformation.
4. AI-Infused Delivery: Using AI Where It Improves the Work
AI-infused delivery does not mean placing AI into every stage of an engagement. It means identifying where AI improves speed, coverage, consistency or responsiveness, while retaining human control over decisions that require judgement and accountability.
AI supports research, data classification, scenario development, forecasting, document review and ongoing monitoring. Consultants then spend more time testing assumptions, interpreting results and connecting the analysis to the client's commercial position.
The commercial demand is visible in the numbers. Accenture reported $2.7 billion in advanced AI revenue during its 2025 financial year, covering generative, agentic and physical AI and excluding data and classical AI, alongside $5.9 billion in related bookings. Those figures show growing demand for AI-related professional services, not that every client is ready for the same level of adoption.
The principle is selective integration. AI belongs where it improves the quality or usefulness of the work, not where it creates the appearance of innovation.
5. Outcome-Based Pricing: Connecting Fees to Value
AI is increasing pressure on pricing models based mainly on hours worked.
Firms can respond by connecting part of their fees to agreed outcomes, such as cost savings, operational improvements, customer satisfaction or implementation progress. Source Global Research has documented AI-enabled consulting firms using combinations of fixed and outcome-based fees rather than relying on one pricing structure.
Outcome-based pricing improves alignment, but it requires clear definitions. Both sides need to agree on what will be measured, when it will be measured and which factors sit outside the consultant's control. Without that clarity, the model creates disputes rather than accountability.
Hybrid pricing therefore remains more practical for many engagements. Fixed fees, subscriptions, time-based work and performance-linked elements combine according to the type of service and the level of influence the firm has over the result.
6. Continuous Transformation: Treating Reinvention as an Operating Discipline
AI capabilities, regulations and client expectations keep changing. A consulting firm cannot redesign its operating model once and assume it will remain suitable.
Continuous transformation means reviewing how tools are used, how work is divided and whether governance processes still reflect current risks. It also requires firms to update skills, methodologies and reusable assets as evidence from engagements accumulates.
Feedback from completed work should improve future delivery, whether that means refining a model, changing a workflow, strengthening a control or removing a tool that does not provide enough value.
The aim is not constant change for its own sake. It is the ability to adjust when the evidence shows the current approach is no longer effective. Applied together, the six principles describe a firm that keeps earning its position rather than defending one it already holds.
Consulting Beyond the Pyramid
The traditional consulting pyramid was built to scale human effort. AI adds a different form of leverage through systems, reusable knowledge and faster analysis. What emerges is not a flatter version of the same firm but a different arrangement of it: flexible teams, AI-supported delivery, governance built into the operating model, and pricing tied more closely to results.
Human judgement stays central. Its position shifts, from producing the analysis to defining the problem, testing the assumptions and standing behind the decision. That is where AI capability, human intent and measurable impact meet, and it is the part of the work no system takes over.
Consulting beyond the pyramid means moving from effort to impact, from fixed hierarchies to flexible networks, and from one-off recommendations to capabilities clients keep using. Firms that treat this as a tooling exercise will find their economics eroding underneath them. Those that rebuild around it will set the terms for what consulting becomes next.
Next in the Series
Adoption Under Pressure: Challenges, Risks and Pushback on AI in Consulting
The next article examines the organisational, ethical and cultural barriers firms face as they adopt AI.
Frequently Asked Questions
How Is AI Changing the Structure of Consulting Firms?
AI is flattening the consulting pyramid by removing the manual research, analysis and synthesis that large junior teams once provided. Firms are replacing fixed staffing tiers with smaller cross-functional groups of strategists, industry specialists, data scientists and risk professionals, assembled around the needs of each engagement and supported by shared data platforms and reusable tools. Hierarchy does not disappear, because it still allocates responsibility and signs off on advice. It stops being the only route through which knowledge and accountability move.
Will AI Replace Management Consultants?
AI will not replace consultants, but it is already automating parts of their work. Information gathering, document review, data classification and initial analysis are increasingly handled by AI systems. Consultants remain responsible for defining the problem, testing assumptions, understanding organisational context and standing behind the decision. PwC's 2026 Global AI Jobs Barometer found that AI-exposed roles are adding tasks that rely on empathy, judgement and creativity faster than less-exposed roles, which suggests the work concentrates rather than disappears.
What Is a Post-Pyramid Consulting Firm?
A post-pyramid consulting firm is a network of intelligence systems orchestrated by humans. It combines consultants, specialists, AI tools, data platforms and governance processes instead of routing every engagement through a fixed hierarchy. Teams form around the problem, and leadership decides which work belongs with consultants, which tasks AI can support and where specialist review is required. Managers and partners still hold their roles. What changes is that seniority is no longer the only mechanism for organising work.
How Will AI Affect Consulting Fees and Billable Hours?
AI reduces the time required for parts of an engagement, which makes hours worked a weaker measure of value. Billable hours will not disappear, particularly in complex programmes requiring implementation and stakeholder management. Firms are combining time-based fees with fixed pricing, subscriptions, managed services and performance-linked payments. Source Global Research has documented AI-enabled consulting firms using combinations of fixed and outcome-based fees rather than a single structure. The direction is hybrid pricing, not one universal replacement.
What Governance Do Consulting Firms Need When Using AI?
Consulting firms need explicit decision rights covering who approves each AI use case, who owns the system, who validates its outputs, who monitors performance and who is accountable when problems occur. The essential controls are transparency about where AI contributed, bias detection, data provenance and protection, documented human oversight and monitoring that continues after deployment. Governance should be proportionate to the decision, because an internal research task does not warrant the same controls as advice affecting employment, credit or access to essential services.
What Happens to Entry-Level Consulting Jobs Because of AI?
Entry-level consulting roles are changing faster than any other tier, because the tasks that historically justified large analyst cohorts are the ones AI handles best. Junior consultants are moving earlier into work involving client contact, assumption testing and interpretation, rather than spending their first years on data gathering and deck production. The pyramid's economic logic weakens when analysis no longer requires proportional headcount, which puts pressure on how firms recruit, train and develop people through the early career stages.
What Is Outcome-Based Pricing in Consulting?
Outcome-based pricing ties part of a consulting fee to agreed results rather than to hours worked. Typical measures include cost savings, operational improvements, customer satisfaction or implementation progress. The model improves alignment between firm and client, but it requires both sides to agree in advance on what will be measured, when, and which factors sit outside the consultant's control. Without that definition it produces disputes rather than accountability, which is why most firms combine it with fixed fees, subscriptions or time-based elements.
How Metheus Can Help
The shift described in this article is not confined to consulting. Any company deciding what to build in-house, what to buy and where human judgement stays essential faces the same question about where AI creates leverage and where it creates exposure.
We work with growth-stage and scaling businesses on market expansion, and that work is data-oriented throughout: feasibility assessment, market entry planning, go-to-market strategy, demand generation and strategic partnerships. AI has changed how quickly we can reach an answer. It has not changed which answers matter. Market size still needs testing against operational fit. Pricing still needs to hold up against a different buyer in a different market. Digital capacity, regulatory conditions and switching behaviour still shape whether an entry works.
What we bring is the judgement layer between analysis and decision. We define what needs proving before capital is committed, test the assumptions that carry the most risk, and translate evidence into go or no-go recommendations leadership can act on.