AI: the next essential building block for institutional portfolios

As part of our «The AI Finance Hub» initiative, it was a great pleasure and honour for me to conduct this interview.

Dr Miró Mitev, Founder & CEO of Smart Wealth Asset Management AG, Zurich, is a pioneer in AI-driven investing. In 2001, he created one of the first fully automated AI investment processes and later launched the world’s first AI forecasting platform adopted by leading financial institutions. He holds a PhD in Economics and lectures on AI in finance, combining deep expertise with innovation.

I highly recommend that professional investors give this interview a read. Thomas J. Caduff, CEO, Fundplat LLC

Dr Mitev, you are considered a pioneer in AI-powered investing. How did you arrive at this approach?

My background is in operations research and artificial intelligence, and I began working on AI-based systems and neural networks during my PhD more than twenty years ago. Even at that stage, it was clear that financial markets were producing far more data than human decision-makers could process objectively or consistently. The AI model, on whose development I worked for more than twenty-five years, was originally developed for internal purposes at Siemens and used in pension management for asset allocation as well as in the treasury for FX hedging. And if the AI algorithm works well enough to be trusted by Siemens engineers, then it must be reliable.

When I later applied this thinking to asset management, particularly in institutional settings such as pension funds and insurance companies, the opportunity became very concrete. These investors operate under strict regulatory and fiduciary constraints, where behavioural mistakes, key-person risk, and unstable teams can materially impact long-term outcomes.

Smart Wealth was founded to address this exact problem. The AI model was originally designed for institutional investors seeking stronger risk-adjusted returns without increasing overall portfolio risk. Over time, the system has evolved through multiple market cycles, learning how behavioural patterns repeat and how disciplined, data-driven decisions can improve outcomes within conservative frameworks.

For a long time, AI was quiet in the investment sector, but now the topic is experiencing a strong upswing. How do you perceive this development?

The renewed attention on AI is understandable, but it is important to separate long-term capability from short-term enthusiasm. Many firms are only now exploring AI because computational breakthroughs and data availability have improved, not because the underlying investment challenges are new or require it.

In institutional investing, particularly for pension funds, the real challenge has never been a lack of technology. It has been the assumption that higher returns must always come with higher risk. Over the past twenty years, Swiss pension funds have delivered stability and trust, but within a narrow return corridor shaped by regulation, demographics, and falling interest rates.

What has changed is that the combination of AI model training over multiple market cycles and quantitative modelling now allows this trade-off to be challenged. AI, when properly designed, can enhance returns without a proportional increase in risk. Years of AI model training have allowed us to capitalise on this sea change. To identify and act upon a trend still requires generational time, data, and experience across multiple market cycles.

Where do you think this dynamic will lead? Would you venture a prediction?

I believe AI will become an essential building block in institutional portfolios, particularly for investors such as pension funds that prioritise liquidity, capital preservation, and long-term resilience.

However, this does not mean a move towards speculative or unconstrained strategies. Any use case of AI in asset management must respect regulatory frameworks and focus on risk-adjusted performance over many years, and must be done through a proprietary model, not a third-party LLM.

Pension funds aim not for maximum returns, but better outcomes within a set risk budget.
Dr Miró Mitev

For pension funds, the objective is not to maximise returns at any cost, but to improve outcomes within a clearly defined risk budget. This is something AI has become a master at achieving.

In that sense, AI will increasingly be used to break the traditional assumption that risk and return are linearly linked. Used thoughtfully, it allows incremental return enhancement, better diversification, and faster recovery from drawdowns, all while remaining aligned with fiduciary responsibilities.

How exactly do you transfer your AI know-how into the practical application of your company?

At Smart Wealth, AI doesn’t play a supporting role; it is the only decision-maker. We define the strategic boundaries, such as eligible asset classes, liquidity requirements, and risk limits, and once those parameters are set, human intervention ends.

The investment universe is deliberately institutional in nature, consisting of highly liquid blue-chip equities, bond ETFs, and commodity ETCs. This ensures daily liquidity and allows the AI to adjust exposure dynamically, including increasing cash positions when early warning signals appear.

By removing human bias, key-person dependence, and emotional decision-making, the system can respond objectively to changing market conditions. Historically, this has resulted in significantly faster recovery after major drawdowns and stronger risk-adjusted returns than traditional benchmark portfolios, while operating well within volatility levels that institutional investors are comfortable with.

What goals have you set for yourself this year?

Our primary goal is to continue demonstrating that the traditional risk-return trade-off does not need to be binary. Swiss pension funds have been forced to choose between stability and return for decades. We believe that well-designed, AI-driven strategies can help change that perception.

This year, we’re focused on working closely with our pension fund investors.
Dr Miró Mitev

This year, our focus is on working closely with the multiple pension funds that have invested in our funds and on continuing to show institutional investors how an AI system that has been running for multiple market cycles can function as a practical component within existing portfolios, enhancing expected returns, improving diversification, and reducing dependence on interest-rate-sensitive assets, all without compromising liquidity or regulatory discipline.

We are not focused on rapid expansion; trust in institutional asset management is built over time, and our priority is to ensure that performance, transparency, and robustness remain aligned with the long-term responsibilities pension funds have to their beneficiaries.