Generative AI's move from experiment to enterprise priority has been faster than almost any technology shift in recent memory. McKinsey's 2024 State of AI survey 1,363 respondents across industries found that 72% of organisations had adopted AI in at least one business function, up from roughly 50% a year earlier. More striking: 65% said their organisations regularly use generative AI, nearly double the share from ten months prior.
That trajectory is what has made generative AI a boardroom conversation rather than an IT one. Leaders are not debating whether technology matters, they are making decisions about where to apply it first, how fast to move, and what foundations need to be in place before scaling.
Why Generative AI Is Different From Previous AI Waves
Earlier waves of enterprise AI predictive analytics, recommendation engines, process automation were largely invisible to the people using the systems. Generative AI is the first wave that puts AI capability directly in the hands of knowledge workers: writing, coding, analysis, customer interaction. That directness is what makes it strategically different.
It means adoption does not depend on a specialist team deploying a model. It spreads through the organisation at the rate that individuals discover it is useful. For business leaders, this creates both an opportunity faster productivity gains across functions and a governance challenge, because the technology is already in use before policies are in place.
Where Leaders Are Seeing Value
According to the McKinsey 2024 data, marketing and sales, product and service development, and IT are the three functions where generative AI is most commonly deployed. The use cases share a common thread: they involve high-volume, language-intensive work where quality and speed both matter.
In practice, the early value is showing up in a few consistent patterns across industries:
Content and Communications at Scale
Drafting, localisation, summarisation, and personalisation tasks that previously required significant human time are being handled or at least started by generative AI. For organisations with large content operations, the throughput gains are real even when every output requires human review before publication.
Code Acceleration
Engineering teams are using AI coding tools to generate boilerplate, suggest completions, and review logic. The productivity gain is not uniform across skill levels; experienced engineers extract more value from AI-generated suggestions than junior engineers do but the direction is consistent.
Knowledge Retrieval and Decision Support
Internal knowledge bases, customer service tooling, and research workflows are being augmented with AI that can surface relevant information faster than a search query. For organisations with large, complex documentation financial services, healthcare, manufacturing this is where some of the earliest measurable ROI is appearing.
The Barriers That Are Slowing Deployment
The gap between "regularly using gen AI" and "generating material EBIT impact from gen AI" is still large. The McKinsey data found that only a small fraction of respondents reported that a meaningful share of their bottom line came from generative AI suggesting that early adoption has not yet translated into production-scale value for most organisations.
Three barriers account for most of that gap:
Data Readiness
Generative AI models are only as useful as the data they can draw on. For most enterprise deployments, the data needed to make AI genuinely useful is spread across legacy systems, inconsistently structured, and governed by policies that were not written with AI in mind. Among high-performing AI adopters in the McKinsey survey, 70% reported data difficulties governance and integration specifically as a primary challenge. Getting the data layer right is a prerequisite, not a follow-on step.
Inaccuracy and Risk
44% of organisations in the McKinsey survey reported experiencing at least one negative consequence from gen AI deployment, with inaccuracy the most common issue. For regulated industries, or for use cases where AI output feeds directly into customer interactions or financial decisions, the risk tolerance for hallucination is low and managing it requires human review workflows, output validation, and ongoing model monitoring.
Governance Gaps
Only 18% of organisations in the survey had established an enterprise-wide council or board for responsible AI governance. The rest are operating with function-level decisions, which creates inconsistency in how risk is assessed, how AI use is disclosed, and what standards outputs are held to. As regulation matures particularly in the EU this gap will become more consequential.
What the Investment Picture Looks Like
Despite the barriers, investment momentum is strong: 67% of respondents to the McKinsey survey expected their organisations to invest more in AI over the next three years. That figure reflects leadership conviction that the technology will matter at scale even for organisations that have not yet seen the returns that justify the conviction.
The investment is going into three places: the AI tools and models themselves; the data and integration infrastructure those models depend on; and the workforce capability needed to use AI effectively. All three are necessary. Organisations that invest in models without fixing the data layer tend to get impressive demos and disappointing production results.
Why Chirpn
The patterns in the adoption data fast uptake, slow value realisation, data barriers, governance gaps match what the Chirpn team sees in client engagements. The organisations that are extracting real value from generative AI are the ones that treated it as an engineering problem before a strategy problem: get the data architecture right, connect AI capability to the systems that need it, and build the governance and monitoring layer in from the start rather than retrofitting it.
Chirpn's AI and ML development practice covers the full stack data architecture, model selection and fine-tuning, integration with enterprise systems, and the production infrastructure that keeps AI accurate over time. AutoPATH, Chirpn's AI-orchestrated delivery framework, provides a structured path from requirements through to production, with AI readiness built into each phase rather than added at the end.
For organisations at the early stages of generative AI deployment figuring out where to start, how to build the data foundation, or how to move from pilot to production, that implementation layer is where the real work happens.
Talk to Chirpn about building a generative AI capability that moves from experiment to production.
Conclusion
The leadership conviction around generative AI is real and well-founded. The technology is already reshaping how knowledge work gets done, and the adoption curve is steeper than for any previous enterprise technology wave. The leaders who are positioning their organisations well are not the ones moving fastest; they are the ones building the data, integration, and governance foundations that will let them scale when the use cases are ready.
That combination of a credible AI strategy, the right technical foundations, and a partner who can build them is what separates the organisations that capture generative AI's value from the ones that accumulate impressive demos.
