In an exclusive interview with Wealth Spark, Bruce Keith (Co-founder & CEO, InvestorAi), the veteran fund manager who built InfraHedge into a market leader, said he is betting that India’s 30 million retail investors are ready to let AI pick their stocks. His startup, InvestorAi—founded in 2018, when most fintech platforms were still automating basic rules—chose the harder path of building proprietary neural networks from scratch.
The platform processes around 150 crore data points daily across the NIFTY 500 to generate one-click investment baskets. The Bengaluru-based firm has scaled to more than 30,000 subscribers across over 20 strategies, with its margin trading fund product recording an 80% win rate soon after launch. Edited excerpts.
Q. InvestorAi is one of the few platforms that has built its own foundational AI model from scratch rather than relying on existing frameworks. What was the original problem or gap that convinced you this much heavy lifting was necessary?
When we started in 2018, there was still a significant portion of AI in financial services that was based on Symbolic AI (basically automating rules) and we felt that this underused the technology. It relied on a human constructing a rule, that by definition could be copied. We had observed that the winners in this market were the high frequency traders and we didn’t feel that the individual investor could properly take advantage of AI.
We set out to construct our own neural networks (more akin to the physiology of a brain than a set of rules) run through a computer vision platform. Had ChatGPT and the other LLMs and reasoning models been available at the current price point, then we may have started from a different place. In retrospect, it has allowed us to create something that would be very difficult to replicate unlike building a model on top of an existing GenAI platform. Our results to date are excellent proof points that we chose the right path.
Q. For a GenZ or millennial investor, “investing as easy as Swiggy or Zepto” is a powerful promise. Can you walk through a real user journey—from opening a broker app to executing a one-click basket—and explain what’s happening under the hood?
One of the reasons that we are different is that we stick to what we are good at and don’t try and force the investor through extra hoops to use the service. As a user journey goes, we are already connected to the investor’s broker so the discovery journey is seamless – you never leave your current brokerage app of choice.
Importantly, the recommendations are executed in a single click (unless the user decides to change them) and again this is inside the broker environment and everything stays in your demat account. Our magic is in picking stocks and creating portfolios and investment baskets – helping users with what to buy, but more importantly what to sell and what to replace them with.
Q. Your AI engine reportedly scans enormous amounts of market data daily
to construct and rebalance equity baskets. In simple terms, how does the model decide what to buy, when to sell, and how often portfolios should be rebalanced?
Every equity basket runs on two to eight different models (depending on the
complexity of the portfolio). Each model looks at 1,024 different features for each
stock, every day for the past 12 years. Across the NIFTY 500, this equates to 150
crore data points in each model, every day. To quote Mark Twain, “History doesn’t
repeat itself, but it rhymes.”
In capital markets this is certainly true and the ability to parse through huge volumes of data in an unbiased way to spot patterns allows our AI to produce consistent outcomes for investors.
Q. You now have over 30,000 subscribers, 20-plus strategies, and strong
ARR growth. Which products or strategy types (e.g., intraday, MTF,
model portfolios) have surprised you the most in terms of adoption and
performance?
Anyone can win once, we’ve all had that lucky break at some point. It is that
consistency, doing it again and again and again that makes a real difference. Over
time our AI has improved and the levels of consistency have increased, however, I
still expect every new product to take a little time to get to this level.
My biggest surprise was probably MTF which had win rates of over 80% straight out of the gate and this hasn’t dropped. The knock on effect for customers is immediately obvious in
their buying and investing behaviours.
Q. InvestorAi’s positioning is about democratising sophisticated wealth-
creation tools for everyday investors. What behavioural patterns do you see in
GenZ and first-time investors, and how has that influenced the way you design
products and interfaces?
Part of the secret is how you combine AI recommendations with markets and
investing experience. I want everyone to be able to experience AI investing and for
them to have a good experience so that they keep coming back. That means that
sometimes you should not have any products available as it is not the right time to
invest.
A great example is our intraday product where we don’t let anyone invest
before 9.30am as the markets are still too volatile and after 11am as there generally
isn’t time to make enough of a return. On other days, we close the window further
and if we don’t see value in the market then we will close it altogether.
It would be easy to always have AI recommendations available, but if new investors lose money then they stop coming back and tire of markets. With new investors of any age, it is not enough to make them aware of the risks, we all have an obligation to help them
know when not to act.
Q.You previously grew InfraHedge into a market leader and have decades of
experience across investment management and technology. What specific
lessons from your earlier ventures have directly shaped how you are building
and scaling InvestorAi?
As you scale up, it is always worth remembering what it was like as a start up.
Specifically, being resource or capital constrained really forces you to make the best
prioritisation decisions. Keeping that in mind, as you grow, means you don’t waste
efforts on side projects that probably should never have started and more importantly
allows you to focus even more on the things that really make a difference.
You should always invest in your team ahead of your growth – this trumps
prioritisation. In the early stages, you need generalists who can turn their hands to
anything and as you grow you should add more specialists. Recruiting people who
are better, more experienced than you is sometimes hard, but is always the best
decision – so long as they are a cultural fit. Building for the long term means that you
should never be bottle neck.
Q. Looking ahead 3–5 years, how do you see AI changing the role of brokers,
fund managers, and retail investors—and where do you want InvestorAi to sit
in that future landscape?
We really are in the early stages of the AI revolution. Think back to the early 90s
when none of us had access to mobile phones, computers, emails and compare that
to the late 2000s – like then we are hardly scratching the surface. There are
profound changes coming and knowledge dependent sectors like asset management
which have been very people heavy will see a transformation that goes beyond
retooling.
In the next 3-5 years, I see customer expectations and demands changing
at a pace similar to what we experienced through the internet revolution and it will be
up to the industry to keep up. We aim to be at the forefront of that, democratising for
an ever increasing investor market both in terms of scale and sophistication.
