A customer service call can quickly become awkward when a voice bot mispronounces a name, cannot understand Hinglish, or pauses too long before replying. Bengaluru-based Arrowhead AI wants to fix that problem with Fletch, its new in-house text-to-speech voice model.

Fletch is designed for India and Southeast Asia, where people often move between languages during a single conversation. It is aimed at banks, NBFCs, insurers, fintech companies and other businesses that make or receive large volumes of customer calls.

What is Arrowhead AI Fletch?

As per their LinkedIn update, Fletch converts written text into spoken audio. In simple terms, it gives an AI calling agent its voice.

Arrowhead AI says Fletch supports 14 languages, including Indian languages as well as Malay and Indonesian. It can also handle code-mixed speech, such as a customer saying, “Mera EMI payment kal due hai,” instead of using only Hindi or English.

That may sound like a small feature, but it matters in real calls. Financial terms, product names, addresses and customer names often appear in English even when the rest of the conversation is in a regional language.

The company says Fletch can start producing audio in under 100 milliseconds and has been tuned for local pronunciation. These are company-reported figures and have not been independently benchmarked.

How does it work?

A voice AI call usually has three main parts. First, speech recognition converts the customer’s spoken words into text. Then, a language model understands the request and decides what to say. Finally, Fletch turns that written reply into speech.

Imagine a customer calling an insurer to ask why a policy renewal payment failed. The AI can listen to the question, check the required information in the company system, explain the next step, and speak back in the customer’s preferred language. If the query becomes too complex, the call can be passed to a human agent.

Arrowhead says it has built several parts of this system in-house, including speech recognition, its language model and the voice layer. The company believes this offers better control over speed, pronunciation and data handling for financial services clients.

The startup behind Fletch

Arrowhead AI was founded in 2022 by Devyani Gupta and Vengadanathan Srinivasan. Gupta is the company’s CEO, while Srinivasan is its CTO.

The startup focuses on voice AI for banks, non-banking financial companies, insurers and fintech platforms. In January 2026, Arrowhead raised USD 3 million in a seed round led by Stellaris Venture Partners. The company has said it operates across India and Southeast Asia.

Fletch is a logical next step for Arrowhead. Rather than depending fully on an outside voice provider, it can tailor its own speech technology for regional languages and industry-specific terms.

Who are the competitors?

Arrowhead enters a busy voice AI market. Global names such as ElevenLabs, Deepgram and Amazon Polly offer speech tools for businesses. Indian AI companies, including Sarvam AI, are also working on language technology for the country’s diverse market.

However, Arrowhead’s approach is more specific. It is not only selling a voice generator. It is offering full calling agents built for sales, collections, renewals and support teams.

That focus could help the company, but it also brings responsibility. For financial calls, businesses must ensure customer consent, data protection, accurate information and an easy way to reach a human agent.