.png){kind=logo}
The increasing reliance on large language models (LLMs) for investment decisions has raised concerns among industry experts, who warn that these tools are being misused. As AI continues to transform the investment landscape, InvestorAi is emerging as a game-changer—bringing sophisticated, market-beating equity strategies to the fingertips of India’s digital investors. Backed by industry veteran Bruce, CEO and Co-founder, the platform combines deep financial expertise with cutting-edge AI technologies like Computer Vision and genetic algorithms to deliver real-time, outperforming stock baskets. With over 34 years in fund management and a proven track record of building market leaders, Bruce is on a mission to democratize smart investing for a new generation.
In an exclusive interview with Analytics Insight, Bruce Keith, Co-Founder and CEO, InvestorAI, sheds light on the limitations of LLMs in investment decision-making and highlights the need for more specialized AI solutions. With a deep understanding of the intersection of AI and finance, Bruce shares his insights on how to harness the power of AI to make more informed investment decisions.
In Conversation with Bruce Keith, Co-founder & CEO of InvestorAi. Analytics Insight explores how AI is revolutionizing wealth management, reshaping the future of AI-driven trading, bridging the gap between human expertise and automated intelligence.
Like any young company trying to break new ground, our story is a mix of inspiration, dedication and perspiration. Our goal from day one has been to democratise wealth management using next gen technologies and we think that we are playing an important role in that.
We started building our AI platform in 2018. It’s not been a smooth ride and the Pandemic forced us to sharpen the offering. Since then we have created a proven track record using our foundational models and have successfully developed products across the investment spectrum.
AI has been part of the investment industry for a long time – I remember dealing with quant shops in the early 2000’s. The difference today is twofold: symbolic AI has largely been overtaken by neural nets and the cost of compute keeps dropping. Twenty years ago AI investing was the preserve of Wall Street hedge funds, today it’s on your phone.
Much of the claimed AI investing that I see is based around LLMs looking at sentiment – while there is some value in this, I only see it as an input. I believe that you need considerably more depth to make a sustained difference in the market.
I also don’t see AI completely replacing the human fund manager – there is a need for both. AI is strongest on quant, ETFs and short term decision making, whereas human equivalents are better at taking long term decisions with less perfect information. From an Indian perspective, I think we will see a real shift from active to passive (ETFs) and quant (AI and systematic strategies) producing better results for investors and reflecting the fund industries in the rest of the world.
As I mentioned above the two co-exist. AI can crunch more data so it really wins the short game, but as humans we are very good at making decisions on less perfect information over a longer period.
This is an industry wide issue. Internally we discuss win, beat and lift rates, but investors remain focussed on returns so we have to create a mix that demonstrates the underlying model performance alongside the investment performance. We publish a set of stats on a regular basis so that potential investors can see exactly how we have performed. Unfortunately the current regulatory framework is restricting some of what can be published and we are working to try and get a framework with clear definitions agreed.
Is AI your brain or a tool to help you? Like any tech, if you abdicate decision making responsibility then you wont necessarily get the outcome you want. AI is fantastic at providing consistency to investment decision making. LLMs generally make it really easy to engage and they don’t judge you on the questions you ask them. However, too many people are using LLMs to make investment decisions – this is not what they are designed for and you might as well just ask google the same thing.
For good reasons, regulators around the world don’t want chatbots making investment recommendations. We have been very clear that our conversational agent, Vani, is designed to help users sift and sort through data and information, but cannot and will not recommend what is good for them.
Vani doesn’t recommend – she is merely there to help people understand what is going on.
Our recommendation models are developed on a proprietary AI platform and foundational models that we have been building since 2018. They are very strong at dealing with significant amounts of data and making short term predictions. We have spent a lot of time in the past 12 months training the models to deal with situations that they haven’t encountered before – in the current environment, that is proving very useful. I am a strong believer that there is good and bad volatility – it’s how you then deal with it.
We have robust processes around releasing, measuring, monitoring and retraining models. This is combined with having challenger models ready at any time and also applying some meta-models that help oversee the whole context. Ultimately no model is infallible and the guardrails are important to catch model drift. Running an AI model without the surrounding infrastructure and governance is irresponsible.
Looking across financial services, there is so much legacy tech that I think many of the traditional players are struggling with where to start. When you overlay the risk of getting it wrong from a regulatory perspective then I think most of the internal AI changes will be aimed at the back and middle office. For this reason we are seeing a large number of existing brokers looking outside to find partnership opportunities that allow them to offer AI investment products that do not increase their own regulatory burden. Given our relative longevity in AI investing we are seeing an increased amount of enquiries.
The regulatory landscape isn’t too different to the rest of the world and is an absolute necessity. We can’t have companies coming in making statements and advising people without the right processes and controls. I really want to see some standardisation of AI performance reporting so that investors can compare and we are working on a framework that we hope will become the industry standard.
As any data scientist will tell you, “It’s all about the data” – there needs to be enough and it must be reliable. We already cover ETFs and have models for crypto (albeit they are not available for public use). The issue with some alternative assets is real time price discovery, in these cases our AI models would not have the same prediction accuracy rates
We are looking to grow 5x this year. We are deploying agentic AI across the entire enterprise to transform our responsiveness, time to market and overall velocity. It has the potential to make scaling much easier and removes some of the traditional points of friction. Exciting times ahead!
