Writing
Essays on payments, fraud, KYC, data protection and how fintech businesses make money. Two a month.
- Credit and MSME finance
A Rs 10 crore guarantee will not fix MSME credit on its own
CGTMSE now backs collateral-free loans up to Rs 10 crore. The bigger gap is the data lenders need to say yes.
- Startups
A founder's guide to collateral-free debt under CGSS
India's credit guarantee for startups now covers loans up to Rs 20 crore. The guarantee lowers the lender's risk, and the lender's standards stay where they were.
- Green finance
Green finance has a last-mile problem
Large solar parks find money easily. The rooftop panel, the e-rickshaw and the efficient motor in a small factory still struggle. Closing that gap is a distribution problem more than a capital problem.
- Value chain finance
Lending along the supply chain: why the anchor matters
Value chain finance lets a small supplier borrow on the strength of a large buyer. India has the rails for it; the hard part is getting buyers to play their role.
- Agritech
Pledge the harvest: warehouse receipts as farm credit
A farmer who sells at harvest sells at the lowest price of the year. An electronic warehouse receipt lets the crop itself become collateral, and a new guarantee makes banks more willing to lend on it.
- Fintech engineering
What a fintech engineer actually needs to know
Writing good code is the entry ticket. The engineers who build lasting financial products also understand money movement, regulation and risk well enough to argue with the business.
- Markets and investing
What three years of trading taught me about building AI products
Markets punish overconfidence within days. AI products learn the same lesson more slowly, and at a higher cost.
- AI
Where should India's next EV chargers go?
Chargers tend to go where land is available. They should go where demand will be, and that is a question data can answer.
- AI
What a win probability curve tells you that a scorecard does not
A scorecard records what happened. A win probability curve shows when the match was actually won, and it works in women's cricket as well as men's.
- Data protection
Dark patterns in financial apps are now a board-level problem
Three regulators now point at the same 13 design tricks. The fix is not a one-time clean-up of the sales page.
- AI
Why an advisor should write code
Advice about technology is better when the advisor has had to make the thing work. Not all of it, and not for every client.
- AI
A demo on dummy data is a promise
Clickable prototypes win attention in financial services. Used carelessly, they also set expectations that real data cannot meet.
- AI
What a machine can and cannot judge in a dark pattern review
Some of the 13 patterns can be checked automatically. Others need a person who understands what the customer was trying to do.
- AI
Pricing a cricketer like an asset
IPL auction prices mostly repeat earlier auctions. Valuing players by the wins they add is closer to how finance values anything.
- Markets and investing
Teaching investing in public when you are a registered research analyst
SEBI has drawn a firm line between education and advice. Being registered does not move the line. It makes it easier to see.
- AI
What RBI's FREE-AI framework asks of banks, in plain terms
The framework favours innovation and asks for governance in return. For most lenders, the hard part is knowing which AI they already use.
- AI
Before you use an LLM to flag risky conversations
Language models can read every call and chat a bank handles. Whether that helps depends on decisions made before the first prompt is written.
- AI
Voice bots for people who do not read: language is the easy part
Speech models now handle many Indian languages well. What decides whether a voice assistant works is everything around the language.
- AI
Government schemes should meet people on WhatsApp
Small business owners miss schemes they qualify for because finding them is hard. The channel they already use every day could fix that, if it can be trusted.
- Data protection
DPDP for product teams: what changes on Monday morning
Most of the obligations under India's data protection law take effect in 2027. Almost all of them land on screens and flows that product teams own.
- Payments
The unit economics missing from every UPI pitch deck
A UPI payment earns the app almost nothing. Any business built on it has to say where the money really comes from, and what each payment costs.
- KYC and fraud
Why fraud models fail at the last mile
Most fraud models detect more than the organisation acts on. The losses happen in the gap between the score and the decision.
- Payments
Open networks are a settlement problem before they are a discovery problem
Protocols like Beckn make it easy for any buyer to find any seller. The hard part is moving money, refunds and disputes between parties who do not know each other.
- KYC and fraud
KYC is a product problem before it is a compliance problem
The rules say what to verify. Whether customers finish the journey depends on choices the rules never mention.