This article is written by Laura Meyer, an Engineer at a leading consultancy, specializing in AI, data science, and DevOps, with extensive experience in GenAI innovation and delivering technical training.
In 2025, AI and data are at the forefront of business transformation, with 89% of organizations expecting a significant impact according to Randy Bean’s report on AI and data leadership in 2025. The shift toward AI in production and at scale is accelerating, and 46.4% of companies are already realizing substantial business value from their data and AI investments. Emerging technologies like generative and agentic AI are reshaping industries, while responsible AI practices are becoming increasingly prioritized. The rise of AI leadership roles signals a deeper integration of data and AI within business strategies. However, challenges such as cultural barriers, talent shortages, and the need for effective operationalization persist. These trends highlight how businesses are navigating the evolving AI landscape to foster innovation, efficiency, and long-term growth in 2025, with a growing emphasis on integrating AI tools into existing processes to drive sustainable value.
1. Reasoning models
In 2025, AI continues to evolve rapidly, but significant challenges remain in fully harnessing its potential for enterprises. AI copilots and search tools have already seen success in helping complete tasks like sentence corrections and answering questions, but reasoning models—those that handle complex, multi-step workflows—are still struggling to meet expectations. While reasoning models can perform standard tasks like generating financial reports, they struggle when faced with unique or non-repetitive scenarios, such as shifting business models. This gap in model accuracy has left enterprises relying more on AI copilots and partial search solutions for the time being.
Take-away: In 2025, AI success will depend on improving prediction and search capabilities, while reasoning models need further enhancement for tackling complex, non-repetitive tasks.
2. Quality of GenAI
The rapid rise of generative AI (GenAI) is driving increased investment in AI and data, with more organizations prioritizing these areas. Investment in AI and data initiatives is projected to rise nearly 20%, from 82.2% in 2024 to 98.4% in 2025. Additionally, the percentage of organizations considering Data & AI a top priority has increased from 87.9% to 90.5%. While many organizations report productivity and efficiency gains from GenAI, its true economic value remains unproven. GenAI’s value typically falls into two categories: reducing costs or generating revenue. On the revenue side, AI-driven tools like sales development representatives (SDRs), enrichment machines, and recommendations can generate substantial sales pipelines, though the quality of these pipelines may be a concern. More commonly, organizations leverage GenAI for cost reduction, especially in repetitive tasks or industries facing labor shortages and urgent hiring needs.
According to a 2025 AI & Data Leadership survey, 58% of data and AI leaders believe GenAI has delivered exponential productivity gains, and 16% say it has freed up knowledge workers from mundane tasks. However, many companies don’t carefully measure these gains or track how workers are spending their extra time. Few studies have documented measurable productivity improvements, with Goldman Sachs showing a modest 20% increase in developer productivity. To gain accurate insights, companies need controlled experiments to assess both productivity and content quality. While GenAI may speed up content creation, this could come at the cost of quality. The true economic impact of GenAI may only be realized with focused measurement and experimentation, and its dramatic effects on productivity may take years. Despite concerns about disinformation, 96.6% believe AI’s benefits outweigh the risks, marking it as the most transformational technology since the Internet.
Take-away: To fully realize GenAI’s economic value, companies must implement controlled experiments to accurately measure productivity gains and content quality, as the true impact may take time to manifest.
Learn more about our Intro to GenAI course, where you’ll immerse yourself in 30 hours of engaging, self-paced content to master prompt engineering and boost your creativity and productivity.
Explore AI tools like ChatGPT, Copilot, and Runway while developing skills in AI content creation, ideation, and personalized sales campaigns. Perfect for beginners, this course includes hands-on practice, 6 exclusive masterclasses, and unlimited access to expert insights to help you leverage generative AI in real-world business scenarios.
3. Agentic AI
Agentic AI, which refers to autonomous programs that collaboratively perform tasks, is emerging as a major trend in 2025. It leverages sophisticated reasoning and iterative planning to autonomously solve complex, multi-step problems, driving productivity and operational improvements across industries. While 37% of IT leaders believe they already have it, and 68% expect it soon, many remain skeptical, viewing it as vendor-driven hype. Despite their potential, current AI agents struggle with accuracy, achieving only 75%-90% in individual tasks.
According to Tomasz Tunguz, a leading futurist and investor, believes that Agentic AI are still far from being ready for deployment. Despite their potential, current AI agents struggle with accuracy, achieving only 75%-90% in individual tasks. When these steps are combined, the overall performance drops significantly, making them unreliable in real-world scenarios. According to Tomasz, AI agents need substantial improvement before they can be used effectively outside of controlled demos or experiments. We believe, early applications will focus on low-risk, structured internal tasks, such as password changes or vacation requests, requiring minimal human oversight. These AI bots will operate in networks but still depend on human review for accuracy. While the workforce impact will remain limited in 2025, opportunities for AI-driven content creation and process automation will increase.
Take-away: While AI copilots excel in specific tasks, Agentic AI requires significant improvements in accuracy before it can be reliably deployed for complex, real-world applications.
4. Unstructured data
GenAI has renewed interest in unstructured data, with 94% of data and AI leaders reporting greater attention to data in 2025. According to an IDC report, only half of an organization’s unstructured data is currently analyzed. With the rise of generative AI, its role in training, fine-tuning, and augmenting AI systems is becoming crucial. As such, much of GenAI’s data is unstructured (e.g., text, images, video), and many organizations are using techniques like retrieval-augmented generation (RAG) to manage and access it. However, managing unstructured data remains labor-intensive, requiring human curation and tools like embeddings and vector databases. While automation may improve over time, substantial human involvement in data preparation will remain essential, especially for tasks like identifying the best sales proposals from many documents.
Take-away: Organizations should prioritize managing unstructured data to enhance AI training, analytics, and platform capabilities, as this data will play a key role in driving business value in 2025.
5. Small data
Tomasz Tunguz predicts that B2B companies will shift towards smaller, proprietary, and open-source AI models due to lower operational costs and improved performance. Small models, like fine-tuned versions of Llama 2, perform better by focusing on specific data, avoiding the errors that larger models with broad use cases often produce. Legal concerns over data rights and the high costs of managed models are also pushing companies toward proprietary solutions, particularly in regulated industries. Meanwhile, B2C companies will continue using off-the-shelf models, benefiting from aggressive price cuts by companies like ChatGPT, which have already reduced prices by 50%.
6. Responsible AI
The emphasis on Responsible AI is growing, with 77.6% of organizations implementing safeguards and frameworks to ensure secure and ethical AI usage in 2025, up from 62.9% in 2024. Despite this, challenges persist, including concerns about misinformation, disinformation, and ethical bias, as well as the need for more AI talent and improved corporate board education on AI. Progress is being made, with more organizations implementing Responsible AI safeguards, but talent shortages remain a significant barrier, as many organizations struggle to recruit the necessary expertise. Additionally, the threat of misinformation and disinformation is now recognized as the most significant risk posed by AI, with its perceived impact increasing over time.
Take-away: Organizations must continue advancing their Responsible AI practices to tackle challenges such as misinformation, ethical bias, and the need for specialized talent, ensuring secure and ethical AI usage.
7. Data and AI leadership progress
Data and AI leadership is expanding rapidly, with more organizations appointing Chief Data Officers (CDOs), Chief Data & Analytics Officers (CDAOs), and Chief Artificial Intelligence Officers (CAIOs). The percentage of organizations with CDO/CDAO roles has risen dramatically from 12% in 2012 to 84.3% in 2025, and 33.1% of organizations now have a CAIO. The focus is shifting toward business value through growth, innovation, and transformation, with 80% of organizations prioritizing these offensive initiatives by 2025. However, there is no consensus on data and AI leadership reporting, as 47.2% of companies still view it as a technology function, despite increasing alignment with business leadership. The CDO role faces challenges with high turnover and short tenures, with 24.1% of CDOs staying under two years. Moreover, the role is often not well-understood by 48.7% of organizations. Despite this, most believe the role is evolving positively, and 70.8% expect the CDO to become a permanent C-Suite position. Overall, in 2025, data and AI leadership will become more integral to driving business success and innovation, with stronger alignment between technology and business strategy.
Take-away: Data and AI leadership roles are expanding, with a growing focus on business value, innovation, and transformation. Stronger alignment between technology and business strategy will drive future success.
8. Operationalizing AI
As AI becomes more production-ready, there’s an increasing realization that the tools themselves aren’t enough. The focus has shifted to how these tools are integrated into existing processes. Data leaders, under pressure to deliver business value, are moving away from experimental, fragmented solutions and towards trusted, end-to-end systems. This shift emphasizes that having the right tools in place is only part of the equation—successful AI adoption hinges on creating well-defined, efficient processes that can be rapidly implemented and scaled. In essence, AI’s future in enterprise will depend not just on better models but on operationalizing them effectively to ensure business value.
Take-away: Successful AI adoption requires not only the right tools but also well-defined, efficient processes. The shift towards integrated, end-to-end solutions will be crucial for delivering long-term business value.
Conclusion
As AI and data continue to drive transformation in 2025, organizations are making significant strides toward leveraging these technologies for innovation, efficiency, and growth. However, AI adoption and transformation remain gradual, with many companies facing cultural challenges, including resistance to change, alignment issues, and skill gaps. These obstacles emphasize the need for a strategic, human-centered approach to AI integration. Despite these challenges, businesses are increasingly recognizing the value of AI, and AI leadership roles are becoming increasingly vital in shaping organizational strategies and driving innovation. The trends of 2025 reflect a continued journey toward unlocking AI’s full potential across industries.
Key takeaways for 2025:
- Focus on improving reasoning models to handle complex, multi-step workflows, as they are still underperforming in real-world, non-repetitive scenarios.
- To fully unlock GenAI’s economic value, companies must prioritize controlled experiments to accurately measure both productivity gains and content quality.
- Agentic AI needs substantial improvements in accuracy before being reliably deployed for complex, real-world applications, with initial use cases focusing on low-risk tasks.
- Managing unstructured data will be crucial for AI training and analytics, as this data plays a pivotal role in driving business value.
- Shifting toward smaller, domain-specific models will improve performance and compliance, as they avoid the errors associated with broader models.
- Organizations must continue advancing their Responsible AI practices to tackle challenges such as misinformation, ethical bias, and talent shortages, ensuring secure and ethical AI usage.
- Data and AI leadership will become more integral to business success, with a growing focus on business value and a stronger alignment between technology and business strategy.
