I build payment products, so the quiet part: huge effort goes into making spending effortless. Almost none goes into saving. $200/month at 7%. Start at 25 and you have about $525k at 65. Start at 35 and it's about $244k. The missing decade costs $281k. saadsalman.org/blog/the-reg...

The regret curve: why we start saving too late

Why financial regret lands later in life, what a ten-year delay costs in rupees (with a compound interest calculator) and four moves that put compounding on your side.

saadsalman.org

An AI agent can book my flight and fill my cart, right up until money has to move. Then it asks me to type my card number. An agentic wallet is three things: an identity, scoped permissions, and a rail. Miss one and you have a liability, not a product. saadsalman.org/blog/agentic...

Agentic wallets: when your money gets an API

AI agents can now book, buy and pay. The missing piece was a wallet they can actually control. Here is the market being built around agentic payments, and where I think it goes.

saadsalman.org

Payments PM interviews are hard to bluff. Ask "what happens when I tap my card at a coffee shop?" and in four minutes you know if someone has touched real money movement. The questions PMs keep asking, and what each tests: saadsalman.org/blog/payment...

A card tap is two events wearing one trench coat. Authorization is a promise: about a second, a hold, no money yet. Clearing and settlement land a day or two later. That gap is why hotel holds linger, refunds take a week and chargebacks exist. saadsalman.org/case-studies...

I see subscriptions from the billing side. Signing up takes one click; leaving takes a gauntlet of discounts, guilt and fake errors. If retention only works because leaving is hard, that's a liability, not a strategy. saadsalman.org/blog/the-sub...

The subscription dilemma: when recurring revenue starts eating the product

How recurring revenue tempts teams into cancellation mazes and other dark patterns, who gets it right, where the FTC click-to-cancel rule stands and a playable maze that proves the point.

saadsalman.org

MUFAP publishes a daily NAV for ~550 Pakistani mutual funds and no API to read it with. So I built one: free, no key, a daily dataset, plus an MCP server so you can ask Claude "which AMCs beat the KSE-100 last year?" saadsalman.org/projects/pak...

Pakistan Mutual Funds API & MCP

Free API, dataset and MCP server for Pakistani mutual fund NAVs, history, returns and performance. Add it to Claude in one line.

saadsalman.org

The fintech demo is a slick app. In MENAP the real value is remittances, salary payouts, SME collections and family money flows. Get those corridors right, cheaply and reliably, and people rely on you. Miss them and nobody will. saadsalman.org/blog/fintech...

My public-speaking club runs on a spreadsheet and a WhatsApp group. Most clubs do, or on web tools older than the smartphone. So I built Rostrum, phone-first: agendas and roles, guest check-in, a live speech timer. In testing with real clubs now. saadsalman.org/projects/ros...

A domino can topple one about 1.5x its size. Grow each by half and the 31st is taller than Everest. I stopped asking how much I got through today. Now I ask: did I knock over the one that sets up tomorrow? saadsalman.org/blog/the-dom...

Every finance app I tried assumed a US bank and a dollar balance. So I built Gramafin for Pakistan: bank accounts, cards, PSX stocks and mutual funds in PKR, rolled into one net worth number. Free, no bank linking. saadsalman.org/projects/gra...

I didn't write the code for my own website. I described it, reviewed what came back, and steered. Next.js, a Sanity CMS, custom SVG illustrations, DNS and SSL, all built through a conversation with Claude Code. saadsalman.org/blog/how-i-b...

How I built this website with Claude Code

From an empty folder to a live site with a CMS, analytics, custom illustrations and its own domain, built end-to-end through one conversation with an AI coding agent.

saadsalman.org

By 2 a.m. I'd crossed off 20 items and moved the product forward zero. Now I start each day with one question: what's the one thing that makes everything else easier or unnecessary? Then I do it before Slack opens. saadsalman.org/blog/the-pow...

The power of one thing

How Gary Keller's focusing question from The One Thing reorganised my workdays as a payments product manager, and the routine I use to pick the one task that makes the rest easier.

saadsalman.org

Every product decision is a bet. Are you betting informed? MIT: data-driven firms are 5–6% more productive. In card programmes that shows up in activation and KYC drop-off. saadsalman.org/case-studies...

Winning your own name is an entity problem, not discovery. Google could already find me. The hard part was standing out from every other Saad Salman. Playbook (open-sourced): saadsalman.org/blog/winning...

I build payment products for a living. Huge talent goes into making spend effortless. Almost nobody engineers that same frictionless magic for your savings rate. Start ugly, start now. Compounding is the one thing money cannot buy back. saadsalman.org/blog/the-reg...

My About page had a complete Person schema graph and 162 words. Google skipped it anyway. Schema does not fix thin content. Word count, query intent, and whether two pages fight the same query come first. saadsalman.org/blog/schema-...

I see subscriptions from the billing side. Once revenue depends on people not leaving, incentives stop aligning with theirs. Cancellation mazes aren't retention. They're a liability. saadsalman.org/blog/the-sub...

That beep after a contactless tap moved zero money. It only reserved a hold and promised the merchant an answer. Authorization is a promise. Settlement is the payment. Most of what cards do that feels weird lives in the gap. saadsalman.org/case-studies...

"Walk me through a card tap at a coffee shop." That's the FizzBuzz of payments PM interviews — looks trivial, filters hard. 15 questions I compared notes on with other payments PMs: saadsalman.org/blog/payment...

In MENAP, cash is the competitor — trusted, universal, free at the point of use. You're not switching cards; you're asking people to trust a screen with money they can hold. What actually changes when you build card programs here: saadsalman.org/blog/fintech...

Fintech in emerging markets plays by different rules

Most fintech writing assumes a US context: cards everywhere, trust by default, one currency. Across MENAP the defaults flip. Here's what actually changes when you build where cash is the incumbent and...

saadsalman.org

An agent can book the flight. The moment money has to move, it stops and asks for your card. An agentic wallet is identity + scoped permissions + a rail. Miss one and you don’t have a product — you have a liability. saadsalman.org/blog/agentic...

Agentic wallets: when your money gets an API

AI agents can now book, buy and pay. The missing piece was a wallet they can actually control. Here is the market being built around agentic payments, and where I think it goes.

saadsalman.org

MUFAP’s “API” returns 500s from anything that isn’t a browser. Their Fund Directory is just a public HTML table. So I scraped that and shipped a free MIT API + a daily dataset with NAV history back to 2022. saadsalman.org/blog/free-ap...

I built a free API for Pakistani mutual fund NAVs (because MUFAP's own one doesn't work)

MUFAP's own JSON endpoints reject any request that isn't a browser. So I scraped their public Fund Directory instead and open-sourced the result — a small, free API serving live NAVs for 500+ Pakistan...

saadsalman.org

A card tap moves zero money. The beep is just authorization, a promise. Clearing runs that night and settlement lands days later. That gap is why refunds are slow, hotel holds linger and chargebacks exist. I built the journey as an interactive ride: saadsalman.org/case-studies/anatomy-of-a-tap

Anatomy of a tap

Ride your payment from the terminal to your bank and back, loop-the-loop included.

saadsalman.org