When CIOs and portfolio managers talk about AI, they are often picturing a much narrower category of technology than what is actually available to them, and that gap shapes how cautiously the industry has moved.
AI has become a strategic imperative in investment management, with significant potential still to be unlocked through its integration into formal investment processes. In conversations with CIOs and portfolio managers, several recurring priorities are shaping how firms approach AI-enabled investing.
We sat down with Stefan Klauser, Co-Founder and CEO of Aisot Technologies, to discuss these priorities and examine some of the assumptions shaping the industry’s approach to AI adoption.
Let's start with what we're hearing from CIOs. While interest in AI is high, many firms remain cautious about integrating it into formal decision-making. What's driving that?
I think a large part of it is definitional. When most investment professionals hear 'AI,' they're picturing large language models, systems like ChatGPT that generate text, summarize documents, and answer questions. And they're right to be cautious about those, because general-purpose LLMs were not designed with financial rigor in mind. They can hallucinate. They don't have a principled relationship with time. They're not built to prevent look-ahead bias.
But that conflation, AI equals LLM, is causing CIOs to apply skepticism about one category of tool to a much broader landscape. What we build at aisot is different. Our approach brings together three things: traditional quantitative finance with established factor models, machine learning to capture non-linear patterns that classical models cannot surface, and LLM-driven news sentiment to process market-moving information in real time. Each layer is trained specifically on financial data, structured around financial principles, and validated against out-of-sample performance. We have built guardrails that eliminate hallucination and look-ahead bias, so the output meets the standards institutional investment processes actually require.
"The skepticism CIOs have about LLMs is valid. What we need is a clearer vocabulary for what else AI can mean in investment management."
How does aisot complement the proprietary data and research that asset managers already consider their core edge?
We see it as synergetic by design. Our approach is to provide the signal infrastructure that sits alongside a firm's proprietary data, not to replace or access it. We process public market data, earnings information, macro indicators, and financial news, and deliver those signals in a format that can be integrated with what a firm already holds internally.
The competitive advantage of proprietary data is real, and we want to protect that. But the research synthesis layer, the work of turning vast amounts of public information into actionable, forward-looking signals, is where most firms are under-resourced. That is where we operate. The two are complementary, not competitive.
"We don't need access to your proprietary data to give your team back the hours they're spending on research they shouldn't have to do manually."
Mid-to-large asset managers often have strong in-house quant teams. What value does aisot add for them?
Quant teams are not looking for us to run their process. They want additional signals they can integrate into what they already have. We work closely with those teams to identify uncorrelated signals that complement their existing factor library and diversify their alpha sources.
What makes our signals distinctive in that context is twofold. First, the news sentiment signals generated by our in-house trained LLMs, the kind of real-time, structured sentiment intelligence that has historically only been accessible to large hedge funds with the resources to build it themselves. On a typical day, news sentiment accounts for 10 to 15 percent of the estimated return forecast. On high-information days, a central bank decision, a major earnings surprise, a geopolitical shock, that contribution can rise to 30 percent. Second, non-linear factor signals created by our machine learning and LLM models go beyond what traditional linear factor models can surface. For a quant team that already has the fundamentals covered, those are the layers that are hardest to replicate internally.
For smaller asset managers and EAMs without a dedicated quant team, what does aisot offer them?
For those firms, the opportunity is different but equally significant. They get access to institutional-grade AI-driven quant infrastructure and signals without needing to build or maintain a quant team themselves. In practice that means they can test and validate investment ideas quickly, accelerating the research process that would otherwise rely on manual work.
We also built a risk layer into the infrastructure that supports portfolio optimization, so smaller teams get risk and opportunity monitoring alongside the signal generation, not as a separate workstream they have to build themselves. Combined with portfolio monitoring and client reporting, that gives smaller teams a level of analytical depth and speed that was previously out of reach. Ultimately it frees up time. Time that can go back into client relationships.