BENGALURU: Brokerage firm Zerodha has invested $5 million in Tijori Finance, a Bengaluru-based market data and research platform that is building AI-driven tools on top of long-term structured and unstructured datasets covering thousands of Indian companies.Tijori founder Siddharth Hegde said the company began by subscribing to standard exchange feeds but quickly realised that fundamental research required information beyond balance sheets and P&L statements. “Structured data is only half the picture,” he told TOI. “A lot of the real insight lies in annual reports, investor presentations and conference calls, but that information is messy, inconsistent and often buried in PDFs. Our entire focus has been breaking down that unstructured data at scale.”This approach led to a deep product integration with Zerodha. “A lot of the data requirements on Zerodha are powered by Tijori today. All their fundamental and mutual fund data comes through us,” Hegde said.One of Tijori’s fastest-growing offerings is Tijori Alerts, which scans roughly 4,000 exchange filings a day and pushes summaries of material announcements – acquisitions, credit rating changes, CFO exits – to users on WhatsApp within seconds. Hegde said this solves a long-standing gap in tracking mid-cap and micro-cap companies. “If you track a small company, news takes days to pick it up. We send relevant updates in under a minute,” he said.Tijori is now scaling Tijori Stack, a suite of AI-powered research tools built on 15 years of historical filings and investor presentations. Its Concall Monitor generates full earnings-call transcripts and AI summaries “within a minute of the call ending,” Hegde said, adding that the system also runs management “consistency checks” across past calls.Hegde noted that accurate outputs depend on the underlying dataset. “Clean data is the crux, 90% of people ignore that. If you feed an LLM bad documents, the output will be bad. We’ve focused on cleaning and structuring the most complex filings and presentations so AI can produce meaningful context,” he said.The company plans to launch several more focused AI modules, with enterprise use cases driving the next phase of growth. “AI is not cheap. Some features work at retail scale, but most advanced capabilities will sit on the enterprise side,” Hegde said.