Arrowhead’s aim is sharp and clear: Build conversational AI that directly impacts business outcomes

What if voice agents engage in natural conversations for as long as 20 minutes, without customers realising they are speaking to AI? What if these AI bots go beyond automation to achieve business outcomes that top-performing human agents are capable of?

This is what Bengaluru-based AI startup Arrowhead aims to achieve.

For decades, banks and NBFCs have relied on large teams of call-centre agents to drive sales, collections, and renewals. But this is expensive, inconsistent, and difficult to scale. 

Training people takes time, attrition is high, and outcomes often vary widely from agent to agent. Bengaluru-based startup Arrowhead believes conversational AI can fundamentally change the scenario. 

Founded in 2022 by Devyani Gupta and Vengadanathan Srinivasan, the company builds AI-powered voice agents designed specifically for the financial services sector—bots that can run long, natural conversations akin to what top-performing human callers are capable of. 

From loan sales to insurance renewals and collections, the platform aims to not only automate high-volume workflows but also improve conversions.

“Indian enterprises have long relied on large human sales teams because labour was considered inexpensive, but this has led to significant inefficiencies, from training and attrition to mis-selling and inconsistent outcomes,” says Devyani Gupta, Co-founder and CEO of Arrowhead.

“We’re building voice AI agents that can handle long, complex sales conversations while delivering meaningfully better conversion outcomes. What’s exciting is how quickly financial institutions are now moving from pilots to full-scale adoption,” she adds. 

Gupta is a Wharton alumna and former consultant at BCG, while Srinivasan is a software engineer with over a decade of experience across companies such as AWS, Rippling, Uber, and Airbnb.

According to the founders, the startup’s voice agents have achieved up to 45–50% higher conversions than human teams in certain deployments, while sustaining conversations that can run for as long as 20 minutes—often without customers realising they are speaking to AI.

Human-like conversations at scale

At the core of Arrowhead’s platform is a proprietary AI architecture encompassing speech-to-text models, large language models, and natural voice synthesis (a combination of in-house and vendor models). 

The platform intelligently routes different parts of a conversation to different AI models based on what a specific moment demands. 

For example, a simple greeting or acknowledgment needs a fast, lightweight model to keep latency low and the conversation feeling natural. While a complex query—like explaining a loan product's terms or handling a KYC objection—requires a more capable reasoning model that can think deeper, even if it takes slightly longer. 

The orchestration layer manages interruptions and timing in real time, so that interactions feel fluid rather than scripted. It also switches between English, Hindi and Hinglish seamlessly. 

All this happens during a single call, taking into account factors such as latency, accuracy, language support, reasoning ability, and response speed. 

Gupta explains, “Think of it like a smart traffic controller. Instead of using one model for everything, which means compromising on speed or intelligence, our orchestration layer makes real-time decisions on which model handles which part of the conversation. 

“This ensures we deliver both low latency and high-quality responses throughout the call, rather than being locked into a one-size-fits-all approach.”

The bots are trained to add pauses, fillers, and tonal variation that mimic natural speech, addressing limitations that make automated voice systems feel rigid.

Enterprise adoption and investor backing

Over the years, Arrowhead has evolved from analysing financial sales calls for compliance and performance risks to building voice AI agents that replicate top-performing human agents. 

Several early analytics clients have adopted its automation platform. 

However, enterprise rollouts are far from being plug-and-play models. For instance, large banks have distinct compliance frameworks and operational workflows. To cater to this, Arrowhead has dedicated engineering and customer success teams to customise deployments. 

Enterprise adoption of Arrowhead’s AI bots has gathered momentum in the last year. The startup says its annual recurring revenue grew fivefold within a few months last year, and every proof-of-concept engagement to date has converted into a live deployment. 

Arrowhead currently serves more than 50 BFSI organisations across India and Southeast Asia, including Bank of Baroda Cards, Aditya Birla Capital, Paytm, Tata 1MG, upGrad, Kissht, Equentis, TurtleMint, InsuranceDekho, and Mudra Fincorp.

The startup positions its conversational AI not as a cost-cutting automation tool but as a performance driver. While the bots operate at much lower costs than human call agents, the bigger value is consistent, predictable performance, says Gupta. For lenders and insurers, this translates to higher revenue outcomes.

“We have achieved up to 45% higher conversion rates and have been able to reach up to 15x more customers than previously,” says Shrey Agarwal, Vice President, Paytm,which uses Arrowhead’s bots to automate customer calls. 

Steady enterprise adoption has led to investor interest. 

Arrowhead recently raised $3 million in a seed funding round led by Stellaris Venture Partners, with participation from angel investors including Kunal Shah (Founder, CRED), Madhusudanan R (Founder, M2P), and several fintech leaders who are also customers of the platform. 

The company plans to use the funding to deepen its BFSI-specific conversation models, expand engineering teams, and build infrastructure capable of supporting tens of thousands of concurrent calls with low latency. Its roadmap also includes emotion-aware voice systems and a broader omnichannel platform spanning calls, chat, and messaging.

For Gupta, the long-term opportunity of conversational AI lies in its ability to directly influence business outcomes rather than functioning purely as an automation layer.

“The next wave of AI adoption in financial services will be defined by results,” she says. “If AI delivers better conversion, consistency, and compliance, adoption becomes inevitable.”

According to Vardhan Dharnidharka, Principal at Stellaris Venture Partners, voice AI for the financial sector in India alone represents a $3-billion market, with less than $50 million penetrated so far, highlighting the early stages the market is in. 

“What’s striking is the speed at which banks and financial institutions are now adopting voice AI, driven by top-down, organisation-wide AI mandates. Arrowhead’s focus on delivering superior conversion outcomes, combined with its strong traction among leading financial institutions, positions it well to capture this opportunity as the market scales,” says Dharnidharka.