Notes on India's financial infrastructure.
Plain-English explainers on payments, identity, lending and fraud, written for people building in Indian BFSI.
How to build an AI calling agent (and when to buy one instead).
Telephony, ASR, LLM, tools and TTS: the architecture, latency budget, Indian-language pitfalls and build vs buy.
AI call centers in India: what changes when an AI answers the phone.
IVR vs AI, which inbound calls to automate, human handover, and the metrics that matter.
WhatsApp AI agent for business in India: how it works, what it costs, what Meta allows.
WhatsApp Business Platform, per-message pricing, the 24-hour window, Meta’s AI rules and BFSI use cases.
AI calling agent in India: how it works, what it costs, where it fits.
How AI voice agents actually work on Indian phone calls, how they are priced, and where they pay back first.
AI agents in banking: real use cases, team by team.
Collections, KYC, service, fraud, lending ops and payments: where AI agents help, and where they should not.
AI telecaller vs human telecaller: an honest comparison.
Cost, scale, quality and compliance side by side, and which calls should stay with people.
Hindi AI voice agents: what it takes to sound right on real Indian calls.
Hinglish, accents, numbers and noisy lines: what makes a Hindi AI calling agent work, and how to test one.
RBI rules on loan recovery calls: timing, harassment and your rights.
Call timing, banned practices, phone-locking limits and how to complain. For borrowers and lenders.
AI voice agents for loan collections in India: what works, what RBI requires, how to choose.
DPD buckets, RBI calling rules (recording from 2027), pricing, and a checklist for choosing a vendor.
Is AI cold calling legal in India? The rules, in plain English.
TRAI number series, consent, DND, RBI calling hours and DPDP: what an AI caller in India must follow.
Why we built Ziya for the whole loan, not just collections.
Notes from building: why collections-only voice AI is backwards, and a live demo to prove it.
Speech analytics for BFSI compliance: why sampling fails.
1–2% QA sampling vs 100% coverage. Disclosures, prohibited language, sentiment.
WhatsApp banking in India: what actually works.
High-volume journeys, platform rules, and the bot-plus-human pattern.
How UPI actually works — a plain-English guide.
NPCI, PSPs, TPAPs, and what happens in the seconds between tapping "pay" and money moving.
KYC in India: the types, and when each applies.
Full KYC, eKYC, V-CIP, CKYC and DigiLocker flows, plus why onboarding funnels leak.