The Bengaluru start-up says its voice agents resolve 70% of borrower queries without human intervention as it targets complex, multi-step calls across financial services.
Karta AI, an agentic artificial intelligence start-up founded by former Razorpay executives, has raised $3mn in seed funding as it tries to automate some of the most complex customer service calls in banking and financial services. The round was led by PeerCapital and Cedar Hill Capital, with participation from Wyser VC, CDM Capital and Seedstars. According to the company, the new capital will be directed towards sales and distribution over the next two years, while also funding product development in areas such as self-learning systems.
The Bengaluru-based company, launched in 2024 by Vijaypal Singh and Abhishek Kushwaha, sells an omnichannel customer communications platform that works across voice, chat, email and WhatsApp. But its main focus is inbound voice calls, where customers often arrive with more than one problem and expect the system to pull information from several enterprise tools before responding. Karta AI says it already handles about 2 million customer conversations each month in more than 25 languages, and counts PhonePe, Mindlabs, Bajaj and Aavas among its paying customers.
Singh told Inc42 that the start-up’s software is designed for what it calls level 2 and level 3 queries, rather than simple frequently asked questions. In practice, that means a borrower might ask for a loan disbursal update and a foreclosure statement in the same call, forcing the system to identify each request, fetch the relevant data and trigger the right workflow. The company says its agents resolve about 70% of borrower queries without human intervention and can cut the cost of handling a query by roughly half. It also says its proof-of-concept deployments have achieved end-to-end automation rates of 70% to 80%.
That positioning reflects a broader trend in enterprise AI, where lenders, insurers and other financial firms are looking for tools that can do more than answer scripted questions. Karta AI says it does not try to replace human staff in every case: high-risk issues such as fraud, or processes without clear standard operating procedures, are still handed to people, and customers can request a human agent at any point. The company’s platform uses a context engine that draws on standard operating procedures, knowledge bases, previous conversations and application programming interface data to keep track of follow-up requests within the same call. It also says it uses models from OpenAI and Google’s Gemini family, alongside open-source systems such as Kimi and Gemma.
Karta AI’s voice product currently supports around 10 to 12 Indian languages, while its chat service supports more than 200. The company says its voice system is built to handle Hindi-English code-switching, regional accents and dialects, although performance varies by language. Because it works in financial services, it has also added compliance checks and guardrails, with agents tested by Karta AI, customers and AI-based testing systems before deployment. The start-up says it plans to build reusable sector-specific capabilities in India before expanding overseas, betting that the same automation model can later be applied to other markets with heavy customer-service workloads.
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.
