In focus: How Weya AI is reimagining enterprise customer support through voice-first AI agents
Call your bank. An AI picks up instantly, recalls your last conversation, and follows up on WhatsApp if the call drops. That's the experience Weya AI is shipping into production for BFSI teams that can't afford missed calls or stale follow-ups.
Founded in December 2024 by Hasan Ali (CEO) with co-founders Atul Singh and Himanshu Tiwari, the Noida-based startup builds memory-driven agents that work across voice, WhatsApp, and email. The goal: keep context, continue the thread, and close the loop-without making customers repeat themselves.
What Weya AI actually does
Weya AI connects to a bank or NBFC's CRM, launches real-time outbound or inbound conversations, and keeps context across channels. If a call drops, it resumes on WhatsApp. If the customer replies later by email, the agent already knows the history.
The platform supports 10 Indian languages and 26 global languages for teams operating in the UAE, Australia, Indonesia, and the US. This makes multilingual, compliant customer support easier to scale without bloating headcount.
Why BFSI first
Speed and consistency move revenue in financial services. If a loan inquiry waits, a competitor closes it. Weya's agents follow up fast, push next steps, and book callbacks while the lead is still warm.
In under a year, the company has worked with Kotak Mahindra Bank, TVS Motors, and Cars24 Australia, among others. The first client, Rupee112 (Gurugram), came on board in month one and helped validate the model for other BFSI teams.
How the tech works (without outside APIs)
Weya AI runs on its in-house stack. It uses small language models (SLMs) fine-tuned for banking and lending use cases and a custom text-to-speech system to keep conversations natural.
The memory layer is the core. It holds conversation state across voice, WhatsApp, and email so the agent can continue a thread, reference past details, and avoid repetitive verification steps. That continuity drives better CSAT and higher lead conversion.
From gym anecdote to enterprise rollout
Before Weya AI, Hasan Ali built Pulp Vision in 2018, an AI consultancy later acquired by a Brazilian firm in 2022. That stint showed him how to ship AI that works at scale and where it breaks in production.
The spark for Weya came from a simple complaint: a friend at an electricity department fielded endless calls asking when service would return. The idea was straightforward-let AI take the calls like a human and reduce repetitive load. Today that idea is serving banking workflows where every missed call costs money.
Business model and traction
Weya AI offers a SaaS model with two options: pay-as-you-go (per usage minute) and enterprise plans (one-time setup plus monthly subscription). Enterprise clients typically spend Rs 3-4 lakh per month; top-tier accounts reach up to Rs 25 lakh.
In the first 10 months, the company reports 100% month-on-month growth across Indian and international accounts. A partnership with ISON, a Dubai-based call centre network, is helping automate frontline calls across MENA, reducing human workload by up to 80%.
The team is 11 people in Noida and plans a Mumbai office by February 2026 to stay closer to BFSI clients.
Competition and the edge
Competitors include Decagon (US), Lori Keet (Australia), and India's Gnani AI. Weya's bet is narrow focus: go deep on BFSI and solve the last-mile issues-CRM sync, language coverage, compliance workflows, and consistent follow-ups-so banks can rely on it day to day.
The hardest part hasn't been demos; it's trust. Getting Kotak Bank on board required proof in live environments, not pitch decks. That culture-ship, measure, iterate-has been central to adoption.
Funding and what's next
Weya AI is bootstrapped, with a friends-and-family round of ~Rs 60 lakh earlier this year. The company is in discussions to raise its first institutional round by December 2025 to grow across APAC and the Middle East.
On product, the team is preparing to launch Workflows-an automation suite handling end-to-end banking use cases from onboarding and verification to post-loan support. According to Dimension Market Research, the agentic AI market is valued at around $7 billion in 2025 and is expected to reach over $93 billion by 2032 (CAGR ~44.6%). Weya plans to triple its current $500,000 global pipeline in the next six months, adding BFSI clients in Southeast Asia and moving into mid-market financial institutions in Australia.
What this means for customer support leaders
- Reduce wait times: Voice-first agents answer instantly and triage intent, then continue the same thread on WhatsApp or email if needed.
- Increase first-contact resolution: The memory layer carries context so customers don't repeat details, cutting AHT and repeat calls.
- Save leads you'd otherwise lose: Fast, automated follow-ups boost contact rates and loan conversions.
- Unify channels: One conversation state across voice, WhatsApp, and email improves continuity and reporting.
- Scale languages without scaling costs: 10 Indian languages and 26 global languages help multinational teams stay consistent.
- Shift agents to higher-value work: With ISON, Weya reports up to 80% reduction in frontline workload; human teams can focus on exceptions and sensitive cases.
How to pilot in 30 days
- Pick 1-2 high-volume intents: loan status, KYC follow-ups, missed call callbacks.
- Connect CRM and set rules: define lead routing, retry logic, and opt-out handling.
- Select language coverage: map customer geographies to supported languages.
- Define success metrics: FCR, AHT, contact rate, conversion rate, SLA adherence, CSAT.
- Run a limited rollout: start with one region or product line; compare against a control group.
- Review weekly: adjust scripts, compliance prompts, and escalation thresholds based on real transcripts.
Key details at a glance
- Channels: Voice, WhatsApp, Email (single conversation memory across all three)
- Tech stack: In-house SLMs for BFSI + custom TTS + memory layer
- Clients cited: Kotak Mahindra Bank, TVS Motors, Cars24 Australia; first client Rupee112
- Geographies: India, UAE, Australia, Indonesia, US
- Pricing: Pay-as-you-go (per minute) or enterprise (setup + monthly); Rs 3-4 lakh typical, up to Rs 25 lakh
- Roadmap: Workflows for onboarding, verification, and post-loan support
Further learning
Explore finance-focused AI tools and role-based upskilling:
Your membership also unlocks:
