Karta AI Bags $3M to Let AI Take Over Customer Conversations

Bengaluru-based Karta AI has raised a $3 Mn seed round (~₹28 Cr), led by PeerCapital and Cedar Hill Capital, to build AI agents that handle the kind of customer calls banks and insurers most dread the long, messy, multi-step ones.

Most voice-AI startups chase outbound calls or simple FAQ bots. Karta is going after something harder: inbound calls where a customer might raise three different issues in one conversation, each requiring the AI to pull data from a different internal system.

That’s a much tougher problem to automate and it’s exactly the gap Karta is betting banks will pay to close.

The Details

• Round size $3 Mn (~₹28 Cr), Seed round
• Lead investors PeerCapital, Cedar Hill Capital
• Other participants Wyser VC, CDM Capital, Seedstars
• Founded 2024, by ex-Razorpay executives Vijaypal Singh and Abhishek Kushwaha
• Valuation Not disclosed
• Sector context Indian voice-AI funding has jumped from $12.9 Mn to $329 Mn year-on-year through August 2026 a roughly 25x surge, signaling heavy investor appetite in this exact category

Not Your Average Chatbot

Karta’s platform runs across voice, chat, email and WhatsApp, but its real focus is voice calls into banks, lenders, fintechs and insurers.

Unlike basic bots built for Level 1 queries balance checks, simple FAQs Karta targets Level 2 and Level 3 requests: the kind involving loan status, foreclosure, or payment failures that need multiple actions across different systems.

Its “predictive context engine” pulls from past conversations, enterprise procedures and API data to decide what to do next. High-risk or unresolved cases get handed off to a human, with full conversation context attached.

The Traction Behind the Pitch

Karta says it now processes ~2 million conversations a month across more than 25 languages, with 10-12 paying clients including PhonePe, Pine Labs, Bajaj and Aavas.

At one home loan client, the company claims its agent resolves roughly 70% of borrower queries end-to-end, with only 30% escalated to human agents.

That automation rate is the number investors will actually care about it’s the difference between a tool enterprises pilot once and one they scale across their call centers.

Small Seed in a Crowded, Fast-Moving Race

Here’s the part worth noting: Karta’s $3 Mn seed is smaller than what some rivals have raised in single rounds. Navana.ai picked up ₹40 Cr in September, and GreyLabs AI raised ₹85 Cr in October both playing in the same voice-AI-for-BFSI space, alongside Desible.ai and Rezolv.

The fresh capital will go toward go-to-market and distribution over the next two years, plus product development in self-learning AI before Karta even considers international markets.

What the Traction Doesn’t Confirm Yet

A 70-80% automation rate sounds strong, but it’s drawn from proof-of-concept projects and a single home loan client not a broad, audited base. In regulated BFSI environments, a bot that fails on the wrong edge case (a fraud dispute, a compliance-sensitive query) carries real reputational cost, not just a bad customer experience. With better-funded rivals already circling the same niche, Karta’s edge won’t be the “agentic AI” label it’ll be whether its automation rate holds up once it’s running across dozens of enterprise clients instead of a handful of pilots.

The Bottom Line

Karta is picking the harder, messier problem in voice AI and in a sector this well-funded, that might be exactly what keeps it from getting drowned out.