Karta AI, an agentic artificial intelligence startup, has closed a $3 million seed funding round. Inc42 reported that PeerCapital and Cedar Hill Capital led the investment, with Wyser VC, CDM Capital, and Seedstars also participating. The company intends to deploy this capital over the next two years to broaden its distribution channels and go-to-market strategy. Additionally, the funds will support product development, specifically focusing on self-learning capabilities within its AI architecture.
Founded in 2024 by Vijaypal Singh and Abhishek Kushwaha, former Razorpay executives, Karta AI provides an omnichannel communication platform covering voice, chat, email, and WhatsApp. According to Inc42, the startup primarily targets complex inbound interactions within the banking, financial services, and insurance sector. These interactions require coordination across multiple enterprise systems rather than simple automated responses. Singh told Inc42 that the platform handles approximately two million customer conversations monthly across more than 25 languages. The company currently serves 10 to 12 paying clients, including PhonePe, Mindlabs, Bajaj, and Aavas.
The platform focuses on level 2 and level 3 queries, which involve interpreting multiple customer intents and executing workflows across different systems. Inc42 noted that Karta AI claims its agents resolve approximately 70% of borrower inquiries end-to-end without human intervention, cutting the cost per query by roughly 50%. These inquiries include loan disbursal updates, account statements, and foreclosure requests. Singh explained that while level 1 queries account for a significant volume of interactions, they represent a smaller portion of total costs, whereas level 2 and 3 queries drive the majority of customer service expenses.
Technologically, Karta AI utilizes a proprietary context engine to manage multi-intent conversations. This system retrieves relevant enterprise data from standard operating procedures, knowledge bases, and API data before generating responses. The startup integrates models from OpenAI and Google’s Gemini family, alongside open-source models like Kimi and Gemma. To accommodate the Indian market, the voice platform supports 10 to 12 Indian languages and is designed to handle code-switching between Hindi and English, as well as regional accents. Compliance measures are built into the platform, with agents undergoing testing by the company’s team, customers, and AI-based systems before deployment.