The education sector has risen to the top among all industries in AI adoption. Presentations noted that a large share of educational institutions already use generative AI, and in students’ everyday lives AI has become a natural part of learning. Education is an industry where AI’s impact is both the greatest and where the need has been identified most clearly.
At the same time, a key challenge emerged: a skills gap. Even though AI use is becoming more common quickly, teachers, students, and organizations feel their understanding and practical skills are not developing at the same pace. This highlights the need for systematic, long-term AI development instead of isolated experiments.
The event also looked ahead to the next development phase: autonomous AI agents. While Copilot-type solutions support individual tasks—such as writing, analysis, or information retrieval—agent-based AI can handle entire processes independently. This fundamentally changes how organizations operate and serve their customers.
“Every employee will have a copilot. Every business process will have an agent.”
The discussion also raised a vision of the future university. According to Gartner research, by 2028 over 70% of learning materials will be produced with the help of generative AI. This is not only about efficiency, but entirely new ways to personalize learning and free up teachers’ time for what matters most. This development is already underway.
Aalto University serves as a concrete example of goal-oriented AI adoption. The presentation described some of the key steps in Aalto’s AI journey: why it has invested in AI, some of the learnings, and some of the ways that AI supports daily work in for both academic and services staff. Aalto’s approach emphasizes strategic consistency, a culture of experimentation, and development driven by practical needs.
One of the key milestones was the Chat with aalto.fi AI solution, in whose development and implementation Modirum Platforms and Microsoft have participated. The goal being to help internal users find answers quickly without needing to search and browse multiple pages. Due to the success of the project, two new chatbots designed for different target groups: Student guide, which serves Aalto students with study-related questions, and Studies bot, which helps prospective students for example with matters related to applying, will be launched in the coming months. Both bots use the same broad source content from Aalto’s website (30,000+ pages), but they are built to meet the needs of different user groups—also taking into account user-specific content restrictions and multilingualism (Finnish, Swedish, English).
The technological renewal of the solution has enabled a transparent, continuously evolving architecture in which information retrieval, orchestration logic, and response generation can be managed and optimized. This way, AI does not remain a “black box,” but instead develops iteratively toward an increasingly reliable and targeted user experience.
A traditional IT project is based on predefined requirements and rules, whereas in an AI project the outcome is inherently probabilistic and can only really be understood through experimentation. That’s why success depends on validating business viability and data suitability early, and on iterative work where quality improves from one cycle to the next. For an AI solution, you must define what “good enough” looks like and measure it systematically, because perfection is rarely achievable in practice. Ultimately, an AI project doesn’t “finish” as a one-off delivery—it requires ongoing monitoring, feedback, and continuous improvement in production.
Sources:
1, 2, 3, 4: Microsoft presentation at AI & Agent Lab
5: Robert Salvén, Aalto University, Presentation at AI & Agent Lab
