Opening a trading account in India now takes minutes. Knowing what to do once you’re in is much harder.
This gap was the focus of Dale Vaz’s keynote at TechSparks 2026. The co-founder and CEO of fintech company Sahi argued that the last decade of fintech solved the issue of access, but not outcomes.
“Everybody here has a broking app,” Vaz told the audience. “But when you think about outcomes, do people really feel confident about what to do with their money? The numbers indicate most of us are still struggling.”
The intent-capability gap
Vaz described a familiar pattern among retail investors. People start investing, then struggle to make sense of their portfolios. A bad day in the market leaves them unsure about whether to buy, sell, or quit altogether.
The cause, he said, is complexity: too much jargon, too much data, and too many tools. Investors are left with two poor choices. They can spend years becoming experts, or they can trust someone promising guaranteed returns.
Regulator-led investor education has helped, Vaz said, but a wide gap remains between people’s intent to grow their money and their capability to do it.
Why generic AI falls short
AI is increasingly pitched as the answer. But Vaz cautioned that general-purpose chatbots carry three serious flaws when money is involved.
The first is hallucination. “A model is confidently wrong. It will never tell you it’s wrong,” he said. That might be acceptable for an essay, but not for financial decisions.
The second is missing context. A chatbot can answer 'should I buy HDFC stock?' without knowing that half your portfolio is already in banking stocks.
The third is accountability. If a general chatbot gives you bad guidance, there is no regulatory protection and no one to call.
Meet Cairo: A purpose-built assistant
Sahi’s answer is Cairo, an agentic AI assistant built into its app and web platform. Vaz called it a “vertical harness” around an AI model. If the model is the engine, the harness is the car.
Cairo combines three things: personal context from a user’s activity on Sahi, 28 expert skills written by Sahi’s research team, and more than 100 tools. Those tools give Cairo “hands and legs," he said. Users can set price alerts, create conditional orders, build custom scanners, or describe a trading strategy in plain language and have Cairo turn it into executable code.
Crucially, the AI model doesn’t do the calculations. It routes requests to Sahi’s backend APIs, which return deterministic results. “When you say add four plus five, the model calls the calculator that we have built,” Vaz explained. This design keeps numbers out of the model’s hands and reduces the risk of hallucination.
A constitution for trust
Sahi built Cairo around a set of rules it calls a constitution. Answers must be grounded in facts, and Cairo is designed to say “I don’t know” when it should. The user always presses the final buy or sell button. Every output must come with evidence a person can check.
“If AI could just magically make money for everybody, the people making the models would have made all the money,” Vaz said. “AI can help augment our skills as investors, but it can never really replace the individual.”
The system is designed to comply with SEBI regulations, with in-country inference, data logging, and full traceability of conversations.
Early results
Cairo has been tested with more than 11,000 traders in a closed user group. It has generated over 2.7 lakh responses at a 99.2% success rate, and 78% of users returned after their first interaction.
Some use cases surprised the team. Many traders treat Cairo as a sounding board for losses and discipline problems. Others talk to it in Gujarati, Tamil, and Marathi.
What Cairo won’t do yet
Asked whether Cairo acts as an advisor, Vaz was direct: it does not. It has strict rules against buy or sell calls. Advisory features would require a separate SEBI investment advisory licence, which Sahi plans to pursue.
Vaz also previewed a Discipline Score. Based on academic research into why retail traders fail, it will flag habits such as overtrading, oversized positions, and revenge trading.
Looking further ahead, Vaz sees AI agents running exchange-approved algorithms for retail investors, and a future where every Indian has “a team of AI-enabled agents” watching their risk, portfolio, and taxes in their own language.
“We are at the first minute of the first day,” he said.