Fresh perspectives on AI in 2025
In 2025, AI stopped being a buzzword and became something else entirely: part of everyday work. Not just in innovation labs or side projects, but inside CRM systems, inboxes, office tools and customer service channels. To make sense of these developments, two of Gen25’s rising experts shared their perspectives: Jessie Guo on the business consulting side and Bart Combee on the technical consulting side. Together, they outline in this article what happened to AI in 2025 and how organisations can move forward into 2026.
What did AI improve in 2025?
Most companies did not struggle with AI ideas in 2025. They struggled with turning those ideas into real value. Where AI did work, Jessie and Bart see three clear themes:
- Knowledge access and enterprise search: a less flashy but very important theme in 2025 was enterprise search, which is essentially a foundational intelligence layer that sits on top of various data sources (think: Salesforce, SharePoint, Google Drive, Teams/Outlook, Slack, etc.), ingests and embeds all enterprise data, and allows for contextual, permission-aware search and actions via natural language. While the underlying concepts have existed for years, the scale and speed of adoption changed this year. More companies started recognising that enterprise search is one of the easiest and most convincing AI business cases to justify. Employees are spending large portions of their week looking for old decks, contracts or documentation, so the return on investment became immediately clear. With models in 2025 becoming far better at summarisation and understanding internal context, enterprise search evolved from a long-standing frustration into the most accessible starting point for organisations beginning their AI journey. This year also marked the rise of AI agents offering similar search capabilities externally, enabling customers to retrieve product details, support information and order updates without needing to wait for human assistance. This broadened enterprise search beyond internal teams and into the customer experience landscape.
- Productivity: 2025 was "the year AI really showed up in everyday work" and no longer only a playground for innovation teams. Employees across many organisations are using tools like Agentforce, Copilot or ChatGPT at scale for information finding or speeding up communication. Beyond drafting emails and documents faster, summarizing meetings and notes and helping developers generate code, these AI tools made it easier to start tasks. LLMs provide the framework for the tasks, makes people feel less blocked and helps them get started right away. As a result, employees spent less time on repetitive tasks and more time for actual thinking. In many cases, people gained back 10 to 20 percent of their week.
- Customer experience: in 2025 AI was also used for help on several customer facing tasks, such as advertising and digital service agents. Those agents helped immensely by answering large amounts of questions, allowing representatives to take their time helping customers. In advertising they helped writers and creators with initial concepts and allowed for faster transfer of visual ideas between employees. However, poor usage of AI in customer facing tasks can lead to reduced perceived brand authenticity resulting in a decreasing brand reputation. Thus, it is important to know the general opinion on AI of your target audience and it is even more important to properly test before launch. AI improves service only when it is thoughtfully integrated into the customer journey. This means that having AI is not the same as getting value from AI.
What can we learn from these AI developments?
2025 being the year in which AI truly came to life, both AI adoption and retention rose significantly. However, research results by MIT showed that around 95 percent of AI investments failed to meet ROI targets. The cause? Unclear business cases and poorly planned rollouts. Although nearly every organisation had AI in place this year, many remained stuck in pilot mode. In her work with customers, Jessie sees that successful AI adoption consistently follows four essential steps:
- Starting with a clear business problem
- Evaluating different AI profiles to match capabilities with needs
- Running structured pilots with regular measurement
- Redesigning workflows supported by change management.
Without these steps, she notes that organisations remain stuck in pilots and never realise enterprise-wide value. The State of AI report of 2025 supports this as companies positioning themselves as “AI-first” grew roughly 50 percent faster in 2025 than those treating AI as a side experiment.
Bart highlighted that 2025 was a technically turbulent year, with innovations arriving so quickly that it’s hard to keep track of all that is happening. Multiple breakthroughs have happened this year, such as the rising popularity of reasoning models. This led to the rise of Agentic AI, where systems take actions rather than simply respond. Jessie acknowledged this pace from the business side. Organisations had to make decisions even as tools evolved rapidly. Features introduced by one vendor were often replicated by others within weeks or months. This meant companies needed strategies and mindsets flexible enough to adjust as capabilities of AI changed. Together, they concluded that selecting the “best” model is no longer the most important decision. Instead, organisations should choose platforms that fit their ecosystem and focus on governance, workflow design and long term strategy.
Looking forward to 2026
Jessie expects 2026 to be defined not by new tools but by scale. More companies will move from AI pilots to AI embedded in standard ways of working. This goes beyond just adding AI to CRM or ERP: it means placing it directly inside the workflows, pages, and actions where people actually do their work. When AI lives in the same environment as the task itself, it reduces context switching and can draw immediate context from surrounding data. As this embedded approach becomes the norm, companies must invest in skills and responsible usage, so employees understand when AI helps and when it doesn’t, and how to use it safely.
Bart, however, sees it differently and believes 2026 will involve both scaling existing solutions and the emergence of new tools. He also added that governance matters just as much as technology. With rapid changes continuing, organisations need frameworks that allow them to pivot quickly as new capabilities and regulations emerge. Both shared views on the uncertainty created by evolving EU regulations. Some companies in Europe hesitate to invest, while organisations in the US and China move much faster. While cultural and political differences also play a role in adoption speed, waiting too long with AI investment and adoption is risky: starting now, responsibly and strategically, positions companies ahead of the curve.
Jessie and Bart represent a new generation engaging with AI not as a future concept but as a practical tool shaping daily work. Their combined view shows that the real story of AI in 2025 was not about hype or fear but about learning, adjusting and discovering where AI genuinely adds value. In 2026, the question will not be whether companies use AI, but how quickly they can learn to use it well.
