Beyond the Assistant: The AI That Decides (and Acts)
From prediction to intervention: challenges and opportunities for banks and personal finance
Artificial intelligence is evolving day by day, often slowly and almost imperceptibly. What starts as a simple assistant can transform into a concrete, autonomous support, acting on our behalf—potentially in the way we would have acted ourselves, or even better.
Recently, there has been a growing number of solutions and projects centered around the concept of agentive AI and its potential impact on how we live, work, and shop.
To better reflect on this evolution and its real-world applications—some of which are already in motion—we need to go back to last fall.
From Assistant to Autonomous Agent
During that time, Sam Altman, CEO of OpenAI, addressed the topic, stating: “We will get there, and in the not-too-distant future.”
Further reinforcing this vision earlier this month was Sarah Frier, CFO of the same organization, in a post on her LinkedIn profile: “Artificial intelligence is evolving rapidly. The next wave will not just bring more advanced chatbots but AI agents capable of operating independently for long periods, solving complex problems in fields such as science, medicine, and much more.”
According to Sarah Frier, reaching Level 3—a category in OpenAI’s internal evaluation scale that classifies AI advancements step by step on a global scale—is not far off. This level enables AI to operate autonomously over extended periods. What does this mean?
Imagine an AI that doesn’t just answer questions or generate images and videos like today’s generative AI but one that can act:
Scheduling a doctor’s appointment when needed
Intervening in financial markets in real-time
Monitoring events and acting proactively to achieve a specific goal based on unfolding circumstances
This kind of evolution requires increasingly powerful hardware and computational capacity, given the broad range of actions AI agents will be able to perform over time and across different environments.
A major game-changer in this landscape is one of the latest announcements from the U.S. administration: the launch of Stargate, an ambitious AI development infrastructure.
Funded by Oracle, SoftBank, and OpenAI, among others, the project aims to build a massive data center by 2029, directly aligning with the rise of agentive AI.
Stargate could unlock an unprecedented level of computing power. Given its declared goal of creating an environment tailored to developing next-generation AI solutions, this data center could become a critical enabler for the next phase of AI agents, accelerating progress in this direction.
A Continuous, Automated, and Informed Support
Among the industries where agentive AI can make a significant impact, banking and payments stand out.
Artificial intelligence is already embedded in financial services. In some cases, it serves as a support tool for financial advisors, offering virtual assistants and predictive analytics. In others—such as Finn, the AI-driven solution from Dutch bank bunq—AI is at the core of the entire business model.
So, what added value can agentive technology bring to financial institutions?
The introduction of AI agents paves the way for a radical shift in how banks operate, offering:
Virtual financial advisors that automatically manage accounts, optimizing interest rates and reducing fees by reallocating funds.
Predictive financial planning, where AI provides tailored saving and investment strategies for each customer.
Banks as the central hub of a new ecosystem, integrating financial, insurance, and even healthcare services through advanced spending behavior analysis.
Banks already possess an invaluable asset: customer data. By analyzing transactions and purchasing behavior, they could deliver highly personalized financial services, redefining the very concept of a bank. The outcome?
A world where AI learns to understand us, acts accordingly, and simplifies financial management. But there’s a catch on the horizon…
Automation vs. Regulation
While agentive technology effectively addresses the growing demand for automation in the financial sector, it also introduces new challenges—the most critical being regulation.
The financial industry is one of the most heavily regulated worldwide. As a result, increasing efforts are being made to navigate the delicate balance between innovation and compliance. A key example is the launch of the AI Lab by the UK’s Financial Conduct Authority (FCA), aimed at studying and shaping AI’s role in financial services. Additionally, the AI Act has placed significant emphasis on transparency, underscoring the importance of accountability in AI-driven financial solutions.
These initiatives reflect the growing need to ensure that AI advancements align with existing legal frameworks, while still enabling technological progress.
And What About Europe?
In Europe, positions on AI regulation remain divided. During the opening of TechArena 2025 in Stockholm—one of Europe's leading tech events—Swedish Prime Minister Ulf Kristersson advocated for less restrictive regulatory frameworks. He warned that without a more flexible approach, Europe risks “becoming a museum.”
Microsoft and the New Frontier of Quantum Computing
Microsoft has announced a major breakthrough in quantum computing: an innovative chip that utilizes a topological superconductor, a material with unique quantum properties. This chip, named "Majorana 1", is designed to form the fundamental building blocks of a quantum computer more powerful than those available today.
Currently, Majorana 1 remains a research prototype and is not yet available on the market, but the details of this discovery have been published in Nature, one of the most prestigious scientific journals. This innovation could mark a significant step toward the development of more reliable and scalable quantum computers.
Quantum computers promise to revolutionize multiple industries, from drug discovery to cybersecurity, enabling solutions to computational problems in minutes that would take traditional supercomputers years to solve. However, despite their enormous potential, the technology is still in its infancy, and experts predict that the first commercially viable models could emerge within the next five years.
Meta’s Submarine Cable
Meta has unveiled "Project Waterworth", a 50,000 km submarine cable designed to enhance AI infrastructure worldwide. This cable will connect five continents, improving global connectivity and supporting the development of new AI applications.
The Extreme Climate of a Distant Planet
For the first time, astronomers have successfully mapped the 3D atmosphere of a planet beyond our Solar System, revealing an unprecedented and extreme climate.
The discovery concerns WASP-121b, also known as Tylos, a scorching gas giant located 900 light-years from Earth.
Using the Very Large Telescope (VLT) of the European Southern Observatory (ESO), researchers have identified powerful winds transporting elements such as iron and titanium, creating complex weather patterns. They observed two distinct atmospheric flows: a jet stream circling the planet's equator at unprecedented speeds, and a deeper current moving gases from the side permanently exposed to its star—where temperatures are blistering—toward the cooler night side.
The essay by Mustafa Suleyman, CEO of Microsoft AI and co-founder of DeepMind, is a must-read for anyone looking to "connect the dots" in the wave of innovation defining this era. Bill Gates described it as “my favorite book on AI because it provides a clear view of both the opportunities and risks (…) offering a unique perspective on the subject.”
OpenAI has introduced its first operational AI agent in the United States, capable of autonomously performing tasks for users such as browsing the web, making restaurant reservations, filling out forms, or ordering groceries. While this is still a first-generation product, it has already generated significant interest, offering an early glimpse into what the future may hold.
Currently, the service is available exclusively to “Pro” users—who pay approximately $200 per month—and is limited to the U.S.. OpenAI may expand access to other countries and subscription tiers in the future, but in Europe, rollout could take longer due to regulatory considerations. The AI agent may eventually be integrated into ChatGPT.
The goal of Operator is to streamline repetitive online tasks, leveraging a built-in browser and a new model called the Computer-Using Agent (CUA). Essentially, it can interact with graphical interfaces and refine its actions over time thanks to its reasoning capabilities.
Our keyword for this post is Agentive Technology
Agentive technologies represent a significant evolution in artificial intelligence: no longer mere assistants, but true digital delegates capable of operating autonomously, coordinating tasks and decisions to optimize processes and resources. Unlike traditional AI, agentive technologies do not just support users—they act on their behalf, processing information and making choices based on perception, reaction, and output.
This process unfolds through well-defined stages:
Perception
The agent collects and analyzes sensory data, drawing from environmental inputs, specific contexts, or a knowledge base. This phase enables AI to understand the operational landscape and adopt informed decision-making strategies.
Reaction
Once the information is processed, the agent determines and executes an appropriate response. This action may be directed toward a system, another agent, or a person, occurring either in real-time or asynchronously.
Output & Feedback
The agent’s action produces a tangible result, whether physical or virtual. This output is continuously assessed and refined through a feedback loop, ensuring ongoing optimization based on user goals.
Disengage
At the end of the process, the agent either disengages or prepares to reactivate in response to new inputs, maintaining a balance between automation and human oversight.
Thanks for being here. Your support is precious for us.
Please subscribe and share this newsletter with your network to help us build our community.