Himanshu’s experience with a ₹6 stock trade led to the creation of FinEd Saathi, an interactive voice-based tutor designed to demystify stock trading, taxes, and investment products for Indian beginners, emphasising clarity and safety in learning.
A small trade in a ₹6 share became the spark for a much larger project after Himanshu realised he did not fully understand where the real cost of buying and selling a stock was appearing in his account. The gap between the trade price and the final loss pushed him to build FinEd Saathi, a voice-led tutor aimed at people who want plain explanations before putting money into Indian markets.
The idea behind the tool is simple: beginners often need more than definitions. A learner asking about an exchange-traded fund, or ETF, may also need to understand charges, settlement, taxes and the way those numbers appear in a broker app. In India, ETF taxation can differ by category and holding period, with equity, debt, gold and international funds each treated differently. That makes a conversational guide useful, especially for first-time investors trying to avoid costly mistakes.
FinEd Saathi was built during a 10-day voice agents challenge and is designed around a spoken back-and-forth rather than a static lesson. The system can explain stocks, mutual funds, systematic investment plans, ETFs, gold, futures and options, initial public offerings and bonds. It is meant to meet users where they are, rather than assuming they already know market vocabulary or tax terms.
Voice is central to that approach. The product supports English, Hindi and mixed-language speech, allowing a user to ask short questions, interrupt and switch languages naturally. The setup uses Deepgram for speech recognition, Gemini for the teaching dialogue and LiveKit for real-time delivery. The developer also says Murf Falcon 2 was chosen for fast text-to-speech, with an Indian conversational voice intended to make the tutor feel patient rather than mechanical.
The system is divided into separate roles. FinEd handles general learning, while TaxEd is a distinct specialist that is only brought in after explicit permission. That separation matters because ETF tax questions can quickly become technical. The teaching flow is designed to stay on education, not personalised advice, and to avoid pretending to know a user’s tax outcome without the evidence needed to support it.
The project also places limits on what it will do with live market data. If a broker access token is missing or expired, the agent is meant to say so rather than inventing a price. Paper trading starts with virtual cash and allows simulated delivery orders, but the browser must confirm a draft before anything is committed. The developer says no real broker order API is used.
That caution extends to safety and privacy. According to the developer, the system refuses requests for real trading orders, does not ask for sensitive credentials such as passwords, PINs, one-time passwords or full account numbers, and keeps analytics limited to anonymous usage totals. For tax questions, the assistant only answers from a packaged set of official rules and stops if the relevant rule is uncertain or out of date.
The result is less a trading app than a guided learning environment. The lesson from the ₹6 share was not merely that brokerage and taxes reduce returns, but that financial products become far less intimidating when the explanation is clear, the sources are visible and the system knows when not to guess. That, Himanshu argues, is what a useful finance assistant should do.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
