The rapid developments in AI continue, with investments at an all-time high. As the year turns, it’s worth taking stock: Where do we stand in AI development, and what can we expect for 2025?
Three selected videos provide an overview, illustrating the current state and future potential of AI with compelling examples.
One of the key predictions for 2025 is the expansion of Agentic AI. These are artificial intelligence systems designed to act autonomously, make decisions, and perform tasks (e.g., booking tickets, shopping) without requiring constant human intervention.
Another major development is near-infinite memory, building on trends observed in 2024. The number of tokens (such as extensive texts) that an AI algorithm can process has steadily increased. In 2025, it’s expected that these limits will largely disappear. This has two significant implications: First, users will be more willing to feed AI systems with data, as the training effect will no longer be lost. Second, this creates a lock-in effect, as switching to other AI systems will come with high costs. This explains why there’s currently a race among major AI players to develop this feature.
What else? In 2024, we saw a remarkable leap in AI-generated videos, and this trend will continue. In the field of robotics, autonomous systems are expected to become increasingly mature.
Highly recommended is the YouTube video “10 Things Coming In 2025 (AI in 2025 Major Predictions)” published on December 29, 2024 (30 minutes, in English) by TheAIGRID. It explores various types of AI agents (Voice Agents, Customer Agents, Deep Research Agents, Project Management Agents, Workflow Engines) and provides a clear picture of advancements in robotics. Watch here:
For a shorter, more technical perspective, check out the 8-minute video by IBM Technology, released on December 23, 2024. It also covers Agentic AI and near-infinite memory while highlighting two additional trends: the coexistence of very large models and very small models in 2025, as well as advancements in human-machine interfaces. Watch here:
A must-watch: The prompt-driven short film “The Heist” by Jason Zada, powered by Google Veo 2. This video provides an impressive demonstration of the current capabilities of AI-generated video content: The Heist (from Jason Zada)
In a recent interview just before we broke for Christmas, Satya Nadella, CEO of Microsoft outlined a bold yet controversial vision for the future of business applications in a world dominated by AI: Compare Microsoft CEO: AI Agents Will Transform SaaS as We Know It.
According to him, the era of SaaS as we know it is coming to an end, giving way to integrated platforms where AI becomes the central driver. This transformation is poised to disrupt traditional tools and workflows, paving the way for a new generation of applications.
For decades, SaaS applications have been indispensable in powering business operations — CRM, HR, ERP systems, project management tools, and collaborative workspaces have all built their dominance on one fundamental premise: they’re essentially CRUD (Create, Read, Update, Delete) databases with business logic layered on top.
But in the AI era, Nadella envisions that this business logic will migrate to an AI layer, disrupting the role of traditional SaaS apps. “The business logic is all going to these AI agents. They’re not going to discriminate between what the backend is — they’ll update multiple databases, and all the logic will be in the AI tier.” This means that AI agents, as intelligent orchestrators, will integrate seamlessly across multiple SaaS platforms, effectively reducing these apps to their barebones function as databases. The “intelligence” part of these systems — the workflows, decision-making, and automation — will shift to the AI agents, collapsing the individual value proposition of isolated SaaS tools.
In this agent-driven era, AI becomes the “brain” of business systems, with SaaS tools evolving into specialized enablers rather than standalone solutions.
