McKinsey has released its State of AI 2025 report, offering one of the clearest snapshots yet of how AI is actually being used inside organizations. The findings show rapid experimentation, uneven execution, and a widening gap between hype and measurable business value.
Most Companies Are “AI-Curious,” Not “AI-Capable”
According to the report, nearly every large organization is experimenting with AI in some form. Eighty-eight percent say they are using AI somewhere in the business, yet two-thirds admit they are still stuck in pilot mode. Adoption has grown quickly, but operational maturity remains low. Many companies have proofs of concept, prototypes, and isolated tests, but few have moved into scaled deployment.
This gap highlights a central challenge: AI investment is rising, but real transformation demands repeatable workflows, reliable data pipelines, and clear ownership. Many businesses are still figuring out how to turn ideas into output.
AI Agents Are the New Interns
One of the most striking developments is the rise of AI agents. Sixty-two percent of organizations are experimenting with them, but only 10 percent have scaled agent-driven work. Early adopters say AI agents can handle repetitive digital tasks, respond to information, and collaborate with human teams.
McKinsey suggests the first wave of companies to operationalize agents will gain a structural advantage, reducing manual workload and accelerating execution.
HubSpot CTO Dharmesh Shah recently commented on the report, noting that “AI agents are the new interns.” He emphasized that companies should experiment and learn quickly, then lean in and apply AI to real workflows rather than waiting on perfection.
High Performers Use AI to Grow, Not Cut Costs
McKinsey’s findings show that top-performing companies look beyond efficiency. Instead of focusing only on automation and cost reductions, they use AI to reinvent workflows, introduce new products, and scale personalization. These companies are treating AI as a growth engine, not simply a savings tool.
The Skills Gap Remains a Major Barrier
The biggest obstacle is talent. Most organizations say they lack the engineering, data, and product skills required to deploy AI reliably at scale. Companies without strong training programs risk fragmented adoption, inconsistent results, and brand-safety challenges.
Why This Matters for Paid Media and Marketing Teams
The McKinsey report aligns with what many marketing organizations are seeing: AI tools are widely used, but not deeply integrated. Most teams are still testing creative automation, forecasting models, and measurement tools, while fewer have fully re-engineered their planning and buying workflows.
This is where structured, domain-specific AI approaches are gaining traction. For example, in our recent article on Micro Model methodology, we show how localized, high-fidelity models can improve paid media performance by analyzing real campaign and spend data instead of generic benchmarks. The upcoming Camphouse AI Planner, built inside Camphouse AI Labs, applies this methodology directly into media plans, reducing manual work and generating automated media mix recommendations.
This type of targeted, operational AI reflects exactly what McKinsey identifies as the next phase of adoption: moving from experimentation to scalable impact rooted in real organizational data.
The Road Ahead
McKinsey points to a future where AI agents, structured data ownership, and integrated workflows will separate leaders from laggards. Shah summarized it simply: the next frontier is not “using AI,” it is “working with AI.”
Companies that continue to pilot without scaling risk falling behind competitors that operationalize AI in planning, execution, and analysis. Those that invest in training, clean data, and agent-based workflows are likely to see the greatest lift in productivity and innovation.
