The Agentic Leap
How AI got down to business in 2025
2025: A year of agentic AI
If 2024 was the year we were all mesmerised by Artificial Intelligence (AI) that could talk, 2025 was the year we put it to work. This was the year of agentic AI. The novelty of generative AI, with its ability to create text and images, gave way to the promise of a more profound and practical evolution: AI systems that don't just respond, but act.
Agentic AI promises a set of proactive, goal-oriented digital doers capable of pursuing complex goals with minimal human supervision. This marked a pivotal transition from AI as a creative novelty to AI as a functional, practical, indispensable tool that can execute multi-step tasks. In 2025, the full potential of agentic solutions began to crystalise, with organisations firming up their vision for how AI can transform their entire business.
The business balancing act: navigating hype, hurdles, and hires
AI technologies are transforming business operations by enhancing customer service through chatbots, strengthening cybersecurity, and streamlining content creation. In sectors like finance and healthcare, AI is used for precise fraud detection and advanced diagnostics, supporting more informed and nuanced decision-making.
AI is transforming the world of market research too. At Ipsos, AI is enabling us to test advertising creative effectiveness more efficiently, boost engagement and empower participants in online communities through using moderator bots, conduct product testing at scale using datasets enhanced with synthetic data, and more besides.
Consumers recognise this transformative potential. The 2025 Ipsos AI Monitor finds 52% of people across 30 countries saying that AI products and services have made significant changes to their lives over the past 3-5 years, with two-thirds anticipating even more profound impacts in the near future.
Looking ahead, agentic AI – which makes decisions and initiates actions – will be crucial in enhancing business functionalities, such as optimising supply chains and proactively maintaining manufacturing equipment. Agentic AI can orchestrate complex workflows by autonomously delegating tasks to both AI and human agents, with minimal human oversight.
Imagine in the not-so-distant future, an AI orchestration agent aids a company with its marketing campaign. Initially, agentic AI analyses market data and trends to formulate a basic campaign strategy. It then outsources graphic design elements to another AI focusing on visual content creation. For more personalised and creative content, the task is outsourced to a skilled, human freelancer on platforms like Fiverr, who introduces unique human sensibilities and cultural nuances to the campaign. The final product is seamlessly integrated by the original agentic AI, ensuring all components align with the company’s branding and objectives.
Ultimately, the promise of AI in business is to achieve greater efficiency, relevancy and personalisation, leading to a competitive advantage in a rapidly evolving digital landscape.
AI takes the reins, human jobs in strain
AI is reshaping the global job market, but perspectives on the implications of this are mixed. Only 31% of people worldwide believe that AI will improve their country's job market, while 35% foresee a negative impact.
Indeed, not all is rosy in the future. The World Economic Forum's 'Future of Jobs Report' indicates that roles involving administrative tasks, clerical work, white-collar functions, consulting, and data analysis might shrink as AI agents have the potential to automate 60% to 70% of working hours. Moreover, the expected creation of new jobs due to AI has not kept pace with slower economic growth. Alarmingly, the same report predicts that AI could displace up to 92 million jobs by 2030.
This shift will force organisations to rethink how roles are defined – moving from task-based descriptions to those centred on the value people create for clients – and to accelerate the upskilling of their workforce. It will also change how talent is sourced, with organisations focusing more on transferable skills, problem-solving ability, and an individual’s capacity to create value alongside AI workflows. As organisations adopt AI at scale, they will also need to guard against “workslop” – low-effort, AI-generated output that erodes productivity and trust – by redefining performance around discernment, quality, and human judgement in using AI responsibly.
AI in marketing: slop or superpower?
So how do people feel about the use of AI to create visuals or videos in advertising? The jury seems split in this case. To some, the use of AI could indicate an opportunity to personalise and amplify the creative idea, but it could also indicate a reason to distrust the brand.
Attitudes towards involving AI in the creative process vary significantly by country and region, with seven in 10 Indonesians and Thais saying this would make them trust brands more. In contrast, just 15% of people say this in Canada and Belgium.
This mirrors a pattern we see more generally when it comes to AI. Anglosphere countries (Australia, Canada, Great Britain, Ireland, and the US) display much higher levels of nervousness than excitement. European countries show lower levels of nervousness – but similarly limited levels of excitement. At the other end of the scale, countries in Asia are much more positive than nervous.
Looking ahead: what comes next?
For 2026 and beyond, businesses must prepare for two critical trends.
First, the development of responsible AI agents requires collaboration and robust governance frameworks.
This manages challenges such as hallucination rates, energy consumption, and ethical concerns.
Secondly, the boundary of innovation is moving toward Autotelic AI (self-improving AI).
This concept, where AI systems can autonomously conduct their own research processes – from literature review to generating research results – is reaching a new level. Such advancements, including OpenAI’s demonstration of significant progress in reasoning benchmarks, suggest that the future lies in autonomous systems capable of generating new scientific and cultural knowledge at an unprecedented pace, ultimately defining the landscape of the coming decade.
Agentic AI is here; advantage goes to leaders who pair autonomy with accountability. Make transparency a basic requirement – disclose AI use (in the case of advertising, 79% expect this), show provenance, keep humans in the loop – and move from pilots to disciplined production to set the pace in 2026.
Further reading
