Mark Kelly
Founder, AI Ireland
In 2025, artificial intelligence (AI) is shifting rapidly from basic tools to advanced autonomous systems.
Leading the autonomous evolution are AI agents — systems that can reason, plan and execute complex tasks independently. According to Gartner’s forecast, by 2028, these agents will handle 15% of day-to-day organisational decisions autonomously. This transition presents vast opportunities alongside challenges.
Importance of robust cybersecurity
While C-suite executives in Ireland anticipate returns on their AI investments, many organisations still struggle to define success metrics. The key is to move beyond generic applications and develop distinctive solutions that deliver measurable value — all while ensuring robust cybersecurity to protect critical data and systems.
The technological foundations behind this transformation are compelling. Innovations in inference time computation now enable AI models to ‘think’ before responding, dramatically enhancing reasoning capabilities without constant retraining. We are witnessing a divergence in model sizes: some systems are expanding towards an astonishing 50 trillion parameters. Others are being refined and miniaturised for efficient use on personal devices. This demands a parallel evolution in cybersecurity, as more sophisticated systems open new avenues for vulnerabilities.
Decision-making transparency and
accountability remain top priorities.
Customer service emerges as prime testing ground
Modern AI systems now tackle complex problem-solving beyond simple ticket routing, supported by what some describe as ‘near infinite memory.’ This ability to maintain detailed customer histories facilitates personalised service at scale — a development that holds particular promise for Irish businesses seeking a competitive edge.
Use cases and requirements for autonomous AI
As founder of AI Ireland, I have observed that the most exciting applications are emerging in healthcare, employee experience and customer service. In healthcare, AI combined with human expertise enhances medical diagnostics, detecting diseases more accurately and supporting clinical decision-making to improve patient care.
Organisations must also craft distinctive AI ‘personalities’ that resonate with their brand while addressing employee concerns about automation. Success will depend on meticulous attention to user experience, seamless workflow integration and proactive cybersecurity measures.
As robots with generalist capabilities powered by embedded large language models become more common, decision-making transparency and accountability remain top priorities. I firmly believe the future is co-intelligence, and AI literacy has become the most essential skill for long-term success.
