One Customer Call Can Hide Five Problems. Karta AI Raised $3 Mn to Solve Them Together

Most banking voice bots are built to answer a narrow question. Karta AI is going after the harder calls — the ones where a customer may ask about a loan disbursal, request a statement and start another service workflow without ending the conversation.

The Bengaluru-based voice AI startup has raised $3 million in seed funding led by PeerCapital and Cedar Hill Capital, with participation from Wyser VC, CDM Capital and Seedstars. The capital will be used to strengthen its automation and multilingual capabilities, expand deployments with financial institutions, and grow its engineering and go-to-market teams.

Founded in 2024 by former Razorpay executives Vijaypal Singh and Abhishek Kushwaha, Karta AI builds AI agents for inbound customer operations across banks, lenders, fintech companies and insurers. The platform connects with enterprise systems so an agent can retrieve information and execute workflows during the same conversation, escalating to a human when required.

That multi-intent capability is central to the product. Instead of treating every question as an isolated interaction, Karta’s context engine uses standard operating procedures, knowledge bases, API data and previous conversations to understand what the customer is trying to achieve across multiple requests. This allows the system to handle more complex L2 and L3 servicing queries rather than focusing only on basic FAQs.

Karta says it currently handles around 2 million customer conversations every month across more than 25 languages. Its inbound servicing agent resolves roughly 70% of borrower queries end-to-end, according to the company, across use cases including disbursal status, account statements, foreclosure statements and property-document releases. These are company-reported performance figures.

The startup is primarily targeting BFSI, where one customer issue can require information from several disconnected systems and where complex servicing calls can be expensive to handle manually. High-risk situations such as fraud, or workflows without clearly defined procedures, are still routed to human agents.

Karta AI’s larger bet is that voice automation becomes more valuable when it stops behaving like a spoken FAQ. The goal is to let one conversation travel through multiple backend systems and come back with an actual resolution.