Over the past few years, Generative AI has emerged as a transformative technology for global enterprises, yet we are only just beginning to grasp the full scale of its potential. Emerging capabilities such as AI agents, industry-specific workflows and retrieval-augmented generation mean the opportunities around innovating with AI have never been more compelling.
A report from Menlo Ventures found that spending on AI hit $13.8bn in 2024 – a sixfold increase from the previous year. As companies across every industry begin to see returns on their initial investments in both AI enhancements to their products and optimization tools for their workforce, they are still only scratching the surface of how Generative AI can drive value.
The report, which surveyed 600 US enterprise leaders, found that the most popular uses of AI include code copilots (51% enterprise adoption), support chatbots (31%), enterprise search and retrieval (28%), data extraction and transformation (27%), and meeting summarization (24%). While these are undeniably strong use cases for Generative AI – and adoption will continue to grow – they represent only a modest portion of what AI can do.
Most enterprises today use AI to augment human workflows rather than as an autonomous solution in its own right. However, this dynamic is set to evolve, with the next 18 months bringing a shift towards more independent AI-driven systems.
Agentic AI and verticalized workflows
Autonomous agents that can operate with limited human oversight represent the next major frontier in Generative AI. Humans perform many repetitive operations – from IT and DevOps to legal and financial – that AI agents currently in development could easily handle. Increasingly powerful reasoning models like OpenAI’s o3 and DeepSeek’s R1 can solve difficult problems and perform multi-step operations on dynamic information.
Additionally, as established companies have rushed to deploy Generative AI solutions, they have been able to rely on their existing user base. Instead of launching entirely new AI products, most features have been enhancements to current offerings. While this has sped up AI adoption, it has also placed major limitations on how customers can implement and leverage the technology.
Already, we are seeing an increasing number of start-ups building domain-specific applications that leverage Generative AI to create verticalized workflows. Industries like healthcare, legal services, financial services, and media and entertainment are especially primed for disruption by these new tools.
RAG and multi-model strategies
One area consistently delivering value for enterprises is retrieval-augmented generation (RAG). This approach enables AI models to retrieve relevant information from external sources such as databases before generating responses, resulting in greater accuracy and richer context.
With large language models (LLMs) now widely available in various sizes and with increasingly expansive context windows, RAG has become the leading method for enhancing Generative AI applications with additional data and deeper insights.
While OpenAI still has a strong market position, Generative AI has proved to offer a very limited moat. Anthropic and Google are both gaining market share, and open-source models from Meta, DeepSeek, and other research teams offer a compelling alternative to proprietary models.
The ease of substituting different models also allows enterprises to pursue a multi-model strategy. Instead of being tied to one provider, they are free to use whatever model best suits their current needs. Many RAG pipelines, for example, use multiple models to retrieve and process data.
For all the opportunities available, companies looking to deploy AI and gain a competitive edge must act fast, especially as demand for AI talent far exceeds supply.
Despite the growth over the last couple of years, it’s clear that Generative AI has only just begun to gather momentum as a driving force behind new paradigms in automation. Businesses that integrate the next generation of tools today will not just keep up – they will set the pace and future-proof their operations.
Stay Ahead of the Financial Curve with Our Latest Fintech News Updates!
Roman Chernin, Co-founder and CBO at Nebius
Roman is a co-founder of Nebius AI, now responsible for the company business strategy. With over 20 years of experience in the technology sector, Roman’s professional career took off at Naumen, a company that builds B2B products, where he started as a software developer, working his way up to head the business division. In 2011, Roman joined Yandex, one of Europe's largest technology companies. Roman initially worked there as the head of Search, a web and mobile search platform, and in 2016 he became the CEO of the Geoservices business unit.
