A Market Built on Labor Is Being Repriced
For decades, the agency and professional services market was built on a simple economic formula: clients paid for access to specialized human labor. Advertising agencies sold creative teams, media planners and production capacity. PR firms sold writers, strategists and relationships with journalists. Software agencies sold developers, project managers and QA engineers. Consulting firms sold analysts, associates, partners and proprietary frameworks. The larger the firm, the stronger the offer appeared. Scale meant more people, more offices, more sector expertise, more production capacity and more ability to handle complex accounts. In a labor-based model, headcount was not just a cost. It was the product.
Artificial intelligence is now putting that model under pressure.
McKinsey estimates that generative AI could add between $2.6 trillion and $4.4 trillion in annual value to the global economy. Around 75% of that potential value is concentrated in four business functions: customer operations, marketing and sales, software engineering, and research and development. These are not distant or abstract categories. They are exactly the areas where agencies, consultancies, IT services firms, digital studios and outsourcing providers have historically earned their fees. The affected markets are enormous. WPP Media forecast global advertising revenue of about $1.08 trillion in 2025, with digital advertising expected to account for 73.2% of total global ad revenue. Gartner estimated worldwide IT spending at $5.43 trillion in 2025, including about $1.69 trillion in IT services. Gartner also estimated that the worldwide consulting services market grew to $397 billion in 2024, driven partly by demand for traditional AI, generative AI and cybersecurity services. IBISWorld uses a broader definition and places the global management consultants market at around $1 trillion in 2024, which shows how large the professional services economy becomes once broader advisory and business services are included. The significance of AI is not that it can write a headline, summarize a report or generate an image. The deeper change is that it reduces the labor required to produce many professional outputs. A first draft, a customer segment analysis, a code prototype, a research synthesis or a campaign concept can now be produced faster and cheaper than before. When production time falls, the traditional relationship between headcount, delivery capacity and revenue begins to weaken. This is the beginning of the great unbundling of the agency market.
The Old Advantage of Scale Is Becoming Less Defensible
Large agencies and consultancies became dominant because scale created real advantages. A global firm could maintain research departments, creative studios, media buying teams, analytics units, legal support, technology partnerships and delivery centers across multiple markets. Smaller firms could be sharper or more specialized, but they usually could not match the operational depth of a global network. AI is narrowing that gap. The evidence is already visible in productivity research. A controlled study of GitHub Copilot found that software developers using the AI coding assistant completed a JavaScript programming task 55.8% faster than developers who worked without it. In customer operations, a National Bureau of Economic Research study of 5,179 customer support agents found that access to a generative AI assistant increased productivity by 14% on average, with the strongest gains among less experienced workers. The same pattern appears in consulting. A Harvard Business School and Boston Consulting Group field experiment involving 758 consultants found that, on tasks within the capability frontier of GPT-4, consultants using AI completed 12.2% more tasks, worked 25.1% faster and delivered significantly higher-quality work. BCG’s own summary of the experiment reported that around 90% of participants improved their performance on creative ideation tasks, while output quality was roughly 40% higher than the control group. These numbers matter because they attack one of the core reasons clients historically hired larger firms: labor capacity. If a 10-person team can now produce output that previously required 15 or 20 people, the economic advantage of maintaining a large delivery bench becomes less decisive. A boutique software agency using AI-assisted development can move faster. A specialized research consultancy can process more material. A small marketing team can generate and test more creative routes. A solo consultant can produce a level of analysis that previously required junior support. This does not mean large firms disappear. Their advantages are still real: brand recognition, procurement access, regulatory compliance, enterprise security, global account management and deep client relationships. But the basis of competition changes. Size alone is no longer enough. In an AI-enabled market, scale must be converted into proprietary data, workflow systems, domain expertise and measurable outcomes. Otherwise, large organizations risk becoming expensive production machines in a world where production is becoming cheaper.
AI Is Not Destroying Demand. It Is Changing What Clients Pay For
One mistake in the AI debate is assuming that if AI makes execution cheaper, the demand for agencies disappears. That is not what the data suggests. Advertising is still growing. Digital channels continue to absorb more marketing spend. IT services remain a trillion-dollar category. Consulting demand is being supported by cybersecurity, AI transformation, cost reduction, operational redesign and regulatory complexity. Gartner’s forecast shows IT services alone approaching $1.69 trillion in 2025, even as clients become more selective about new spending. The problem for agencies is not falling demand. The problem is where value is moving. Clients are less willing to pay premium rates for tasks they believe AI can accelerate. Basic copywriting, routine SEO content, simple design adaptation, standard desk research, first-draft presentation writing, generic social media calendars and low-complexity development work are becoming easier to compare and easier to substitute. These services do not disappear, but they become harder to price as premium work. At the same time, demand rises for work that AI does not solve by itself: strategic positioning, brand architecture, market entry decisions, reputation management, regulatory-sensitive communication, product strategy, complex transformation, data governance, AI implementation and change management. This explains why the most advanced agencies are trying to move from “we produce things for you” to “we help you operate better.” WPP now describes WPP Open as an agentic marketing platform that connects strategy, creative, media and production through AI agents. Its public materials claim measurable outcomes from client pilots, including 14 extra hours back each week for a team of four and a reduction of strategy and creative development time from four weeks to three hours in one global technology-brand example. That is not traditional agency language. It is software-platform language.
The Boutique Agency Paradox: More Capability, More Competition
AI appears to favor small firms. In many ways, it does. A boutique agency can now use tools for research, analytics, copywriting, design exploration, automation, coding and workflow management that would have required far more resources only a few years ago. The cost of launching a professional services business has fallen. But that creates a paradox: AI increases the capability of small agencies while also increasing the number of competitors. The World Bank estimates that there are between 154 million and 435 million online gig workers globally, representing roughly 4.4% to 12.5% of the global labor force. Demand for online gig work rose 41% between 2016 and the first quarter of 2023. The same World Bank research found that almost 60% of firms in poorer countries reported increased outsourcing to gig workers, showing that distributed professional work is no longer a fringe phenomenon. AI adds another layer to this trend. Freelancers can offer broader services. Consultants can move into adjacent categories. Designers can use AI to produce more concepts. Developers can prototype faster. In-house marketers can create assets that previously required agency support. New agencies can launch without large fixed costs. The result is not simply “small agencies win.” The result is that supply increases faster than differentiation. A boutique agency that once competed against local firms now competes against global freelancers, AI-enabled consultants, niche studios, in-house teams and self-service platforms. Producing work becomes easier. Winning trust, proving expertise and maintaining margins become harder. This is why the agency market is fragmenting. AI gives small teams more leverage, but it also makes generic services easier to copy. The winners are not small firms in general. The winners are small firms with a clear category, deep sector knowledge, proprietary processes or unusually strong client relationships.
In-House Teams Are Becoming Stronger Competitors
The agency market is also being unbundled from the client side. For years, brands have been building in-house marketing capabilities to reduce cost, improve speed and keep data closer to the business. The Association of National Advertisers reported that 82% of its members had in-house agencies in 2023, up from 78% in 2018, 58% in 2013 and 42% in 2008. This matters because AI increases the power of those in-house teams. A brand team that already understands the company, product, customers and internal politics can now use AI to create briefs, generate content variations, summarize research, localize assets, analyze campaign performance and produce first-draft creative work. That does not eliminate the need for external agencies, but it changes what external agencies are hired to do. The old agency promise was: “We have the people and tools you do not have.” The new agency promise must be: “We can solve problems your internal team cannot solve alone.” That is a much higher bar. It also means that agencies will be pulled toward more specialized and strategic work. A brand may not need an external agency to produce 50 social captions. But it may still need help repositioning after a crisis, entering a new market, launching a complex B2B product, redesigning its marketing operating model or building an AI governance framework. The agency relationship shifts from outsourced production to external expertise.
Platforms Are Moving Into the Agency Value Chain
The unbundling of agencies is not only being driven by AI tools. It is also being driven by platforms. Google, Meta, WPP, freelance marketplaces and enterprise software providers are all moving into parts of the workflow that agencies once controlled. Google’s Performance Max uses Google AI across channels to identify conversion opportunities, optimize budget and generate or suggest text, image, logo and video assets from a company’s website. Google’s own documentation says advertisers can use generative AI inside Performance Max to create asset groups “with a few clicks,” while remaining able to review and discard generated assets. Meta is moving in the same direction. Reuters reported that Meta aims to let brands fully create and target advertisements with AI tools by the end of 2026, including image, video, text generation and targeting based on advertiser inputs such as product image and budget. WPP has also moved beyond the traditional agency model. Reuters reported that WPP launched WPP Open Pro to let brands plan, create and publish campaigns using its AI-powered marketing platform, including smaller brands that do not use full-service agencies. This is a major structural shift. Platforms are absorbing execution. They are making campaign planning, asset generation, media optimization and publishing more self-service. That does not make agencies irrelevant, but it compresses the value of routine execution. The same pattern happened in other industries. Uber unbundled transportation companies into drivers, algorithms and demand aggregation. Airbnb unbundled hotels into hosts, listings and platform trust. Shopify unbundled retail infrastructure. AI may do something similar to agencies: separate strategy, production, talent, workflow and distribution into different layers. Agencies that once sold bundled services may need to compete as specialists, platform operators, AI implementation partners or strategic advisors.
Holding Companies Are Responding With Consolidation and AI Infrastructure
Large agency groups understand the threat. Their response has been to invest heavily in AI, data and platform infrastructure. Publicis Groupe announced a €300 million AI investment plan over three years and introduced CoreAI as an “intelligent system” designed to provide AI capabilities across the company. Publicis later reported that it ended 2024 as the world’s largest advertising group, with 5.8% organic growth for the full year, an 18% operating margin and around 114,000 employees across more than 100 countries. Reuters reported in 2025 that Publicis said 73% of its operations were AI-powered and that the company had invested €12 billion in data, technology and AI since 2015. WPP has built WPP Open as an AI-powered marketing operating system. In 2024, WPP and Google Cloud announced a collaboration integrating Google’s Gemini models into WPP Open, which WPP said was already used by more than 35,000 of its people and adopted by clients including Coca-Cola, L’Oréal and Nestlé. Accenture is making similar moves in consulting and technology services. The company announced a $3 billioninvestment over three years in its Data & AI practice in 2023, with plans to double AI talent to 80,000 people. In fiscal 2025, Accenture reported $5.9 billion in generative AI new bookings and annual revenue of $69.7 billion. The agency holding company market is also consolidating. Omnicom announced a stock-for-stock acquisition of Interpublic Group in December 2024. Reuters described the deal as worth about $13.25 billion, creating a combined company with more than $25 billion in revenue and projected annual cost savings of $750 million. Omnicom later announced its go-forward strategy after completing the acquisition of Interpublic in November 2025, positioning the combined company around connected capabilities powered by Omni, its intelligence platform. These moves show that large players are not standing still. They are trying to convert scale into proprietary data, AI platforms, automation systems and integrated workflows. But consolidation also signals pressure. When execution becomes cheaper and clients demand measurable outcomes, large firms need either stronger technology leverage or lower cost structures.
AI Creates Productivity, But Also Quality Risk
The AI transformation is not a simple story of efficiency. The strongest evidence shows that AI can improve productivity dramatically on some tasks while damaging performance on others. The BCG-Harvard experiment is important here because it introduced the idea of a “jagged technological frontier.” On tasks inside the frontier, AI improved speed, output and completion rates. But on a business problem-solving task outside the tool’s competence, BCG reported that participants using GPT-4 performed 23% worse than those who did not use the tool. That finding has major implications for agencies. AI can help generate options, summarize information, draft content, accelerate coding and automate repetitive work. But it can also produce confident errors, shallow analysis, generic positioning, weak strategic logic and brand-inappropriate recommendations. The more complex the client problem, the more important human judgment becomes. This is where premium agencies can defend value. Not by pretending AI does not exist, but by building systems that combine AI speed with expert verification. In regulated sectors such as healthcare, financial services, cybersecurity, climate technology and government communications, the risk of being wrong is high. Clients do not only need faster output. They need defensible decisions. That is why the future agency model is not “humans versus AI.” It is human experts using AI inside controlled processes, with clear accountability for quality, accuracy, compliance and business impact.
The Labor Model Is Already Changing
The impact on agency labor is no longer theoretical. The UK’s IPA Agency Census 2025 found that member agencies employed 24,963 people as of September 1, 2025, down 6.8% from the previous year. Creative and other non-media agencies were hit harder, with employment falling 14.3%. Employees aged 25 and under declined 19.2%, while vacancies fell 40.8% across all seniority levels. The same census found that 88.3% of agencies said AI was having a considerable impact on how they work. 8% of agencies had reduced workforce in the previous 12 months as a direct result of AI, and 24% expected to do so in the next 12 months. Among creative and other non-media agencies, the expected figure was 30%. This does not prove that AI is the only cause of job reductions. The agency market is also affected by client budget pressure, procurement changes, media fragmentation, macroeconomic uncertainty and the shift to in-house teams. But AI is clearly becoming part of workforce planning. The same pattern is visible in IT services. In June 2026, Reuters reported that Indian IT stocks had their worst day in four months because of concerns that AI could disrupt traditional software services. Tata Consultancy Services fell 9%, the Nifty IT index dropped 5.8%, and analysts raised concerns about AI reducing demand for legacy outsourcing. The signal is clear: investors, clients and agencies are all reassessing the value of labor-intensive service models.
Client Expectations Are Moving From Deliverables to Outcomes
A report, a campaign, a website or a deck is no longer enough. Clients want to know whether the work increases revenue, reduces acquisition costs, improves conversion, protects reputation, accelerates product launch, reduces support volume, improves retention or strengthens competitive position. This shift is visible in enterprise AI adoption data. McKinsey’s 2025 State of AI survey found that 88% of respondents said their organizations were using AI in at least one business function, but nearly two-thirds had not yet begun scaling AI across the enterprise. Only 39% reported EBIT impact at the enterprise level. This creates a major opportunity for agencies and consultancies. Many companies are experimenting with AI, but they are not yet capturing enterprise-level value. They do not need another vendor promising “AI-powered content.” They need partners who can redesign workflows, define use cases, manage risk, train teams, integrate tools, measure impact and connect AI adoption to business performance. This is where professional services can move up the value chain. The agency of the future may earn less from producing individual deliverables and more from building systems: campaign operating systems, content supply chains, customer intelligence workflows, automated reporting environments, AI governance processes and performance measurement models.
What Will Be Commoditized — and What Will Become More Valuable
The most exposed services are those with standardized inputs, repeatable processes and low differentiation. These include basic content production, generic SEO articles, simple ad variations, first-draft research summaries, standard reporting, low-complexity design adaptation, simple landing pages and routine QA tasks. The more defensible services share different characteristics. They require context, judgment, relationships, accountability or domain expertise. These include strategic positioning, crisis communications, executive advisory, category strategy, regulatory communication, enterprise transformation, complex B2B marketing, AI implementation, data governance, high-stakes creative direction and industry-specific consulting. The difference is not “creative versus technical.” AI affects both. The difference is whether the work is primarily execution or judgment. A generic social post can be generated. A brand repositioning after a reputational crisis cannot be safely delegated to a model. A simple product description can be automated. A go-to-market strategy for a cybersecurity company selling into regulated enterprises requires domain expertise. A basic dashboard can be produced quickly. Knowing which metric actually matters to the business requires experience. This is the new agency economics: execution becomes cheaper, but interpretation becomes more valuable.
The Next Agency Model: Smaller, More Specialized, More System-Based
The most likely outcome is not the disappearance of agencies. It is the breakup of the traditional bundled agency model. Some clients will use platforms for routine execution. Some will rely on in-house teams for daily content and media operations. Some will hire freelancers through digital marketplaces. Some will retain large holding companies for global scale, compliance and integrated services. Others will turn to boutique specialists for high-value strategic problems.
- The market becomes more fragmented but also more specialized.
- Successful agencies will have to answer five questions more clearly than before:
- First, what do we know that a client cannot easily get from an AI tool?
- Second, what process, data or methodology makes our work defensible?
- Third, which part of the client’s business outcome do we directly improve?
- Fourth, where do we use AI to reduce cost without reducing quality?
- Fifth, why should a client choose us instead of doing the work in-house or through a platform?
The agencies that cannot answer these questions will be forced into price competition. The agencies that can answer them will become more valuable. By 2030, the strongest agencies may not be the ones with the largest headcount. They may be the ones that best combine human judgment, sector expertise, proprietary workflows, AI-enabled execution and measurable business results. AI is not eliminating professional services. It is redistributing value inside them. The old agency model sold labor. The next agency model will sell judgment, systems and outcomes.