The numbers Brief Episode map Timeline Takeaways Charts Quotes Clips Glossary Go next Transcript

Episode 82 · Fintech · 48 min

From fifty lakh to ten thousand

India's bond market is older than its equity market and, globally, bigger. Until recently it ran on phone calls between institutions dealing in ₹50 lakh multiples. GoldenPi's claim is that the barrier was never retail appetite but lot size: cut the ticket to ₹10,000, put ₹4,000 crore of AAA-to-A paper on a screen every day, and a voice market becomes a marketplace.

Savings & wealth Payments & fintech Product strategy

Abhijit Roy and Samar Prathihar

Co-founders, CEO and CTO, GoldenPi · with Vishal Krishna

48:23

Said in this episode

▶ 10:23

₹10,000

Minimum ticket for a bond on GoldenPi

The entry size retail investors can buy at, against a market lot that historically ran ₹10 lakh to ₹50 lakh.

▶ 16:55

₹4,000 cr

Bond inventory displayed daily

500–600 bonds are available through partner financial institutions, of which 100–150 AAA-to-A rated papers are shown each day.

▶ 8:49

₹50 lakh+

The old minimum trade size

Institutions were trading minimums of ₹50 lakh and multiples of it, the wall that killed GoldenPi's ₹1 lakh portfolio algorithm.

▶ 20:31

8.5% vs 5.5%

AAA bond against the best FD

As stated on air: FDs at 5.5–6%, a ten-year government bond at 7.3%, AAA bonds 7.8–8.5%, AA 9–10.5%, A around 11%.

▶ 15:31

5 trades/day

An institutional bond desk's volume

Roughly five trades a day of ₹50–100 crore each, negotiated over phone calls, against the thousands or lakhs of small tickets retail generates.

▶ 40:23

~$100 tn

Size of the global bond market

Against $65–70 trillion of global equity; India's government securities market is put at about $1.2 trillion and the US at $30–40 trillion.

The brief

The argument in sixty seconds

The founders' claim is that Indian retail never rejected bonds. It was never let in. The bond market predates equities (they trace it to a Mesopotamian seal now sitting in a US museum) and is the largest financial market on earth at roughly $100 trillion against $65–70 trillion of global equity, yet it ran as an institutional game: about five trades a day of ₹50–100 crore each, negotiated over the phone, with minimum lots of ₹10 lakh to ₹50 lakh. Two engineers who met in college, later shared a desk building test rigs for routers and switches, and resigned on the same day in 2016 burned through three or four flips of the model: an exchange route that failed because the bond market simply isn't on the exchange, then a portfolio algorithm that worked until real inventory turned out not to exist at ₹1 lakh. What finally moved the financial institutions was arithmetic: aggregated retail is bigger than the corporate desk. The unlock was a ₹10,000 ticket, 500–600 bonds sourced from partners with 100–150 AAA-to-A papers worth about ₹4,000 crore displayed daily, and a complex rule-based system standing in for the phone call. Nithin Kamath, first approached only for an API, became the backer. The stakes: with the best fixed deposits at 5.5–6% and AAA bonds at 7.8–8.5%, the gap is roughly three points a year, and the harder problem, thin secondary liquidity, GoldenPi still solves by hand, finding an institution willing to buy your paper back.

Worth your time if you are

Savers whose entire fixed income sits in bank FDs

First-time investors who own equities but never a bond

Founders digitising a voice-driven, institutional market

Engineers weighing a fintech job against an e-commerce one

Investors watching yields spike and bond prices fall

Episode map

Where the conversation travels

Every block is a chapter, coloured by what it's about. Click any of it to jump straight to that minute on YouTube.

Savings & wealthPayments & fintechProduct strategyFounder journeyData & digitisationCredit & lending

01Cold open: peace of mind, priced at ₹10,000 0:00 Vishal opens on a market he has always wanted access to, and the founders define GoldenPi as a fixed-income platform whose real product is peace of mind (the portion of a portfolio whose value doesn't give you heartburn) entering at as little as ₹10,000. 02One project, sixteen years, two resignations 2:06 They met in college through a common friend, became colleagues at a second job logging 14-hour days on test infrastructure for routers and switches, and after a failed 2006-07 attempt found themselves in 2016 in the same Bengaluru tech park and the same lane, then resigned the same day. 03Flip one: the exchange with no bond market 5:45 The core problem was fixed on day one but the route wasn't: as techies new to the field they first tried to route retail bond buying through the exchange, only to find the bond market doesn't live there. 04Flip two: a good algorithm, no inventory 8:00 The second attempt built a portfolio algorithm around a ₹1 lakh ticket and died on real data (sellers traded in minimums of ₹50 lakh and multiples of it) so the founders went to financial institutions with the argument that aggregated retail would exceed what the corporate desk does. 05Ownership versus a loan 10:07 The primer that first-time investors need: equity is taking ownership and riding the highs and lows, while a bond is lending the company money against a contractual obligation to pay, which is lower capital risk by construction. 06The letter that reached Nithin Kamath 11:56 Friends and family took the first leap of faith, but the champion arrived when a note asking Nithin Kamath's team for integration APIs turned into an invitation, an intense conversation about the fixed-income market, and his backing. 07A phone-call market meets retail scale 13:15 Build for scale from day one, use what open source gives you, because a hundreds-of-years-old institutional market where five trades of ₹50–100 crore a day are negotiated by voice cannot process the thousands or lakhs of small tickets retail generates. 08500 bonds, ₹4,000 crore on display 16:20 No mutual funds and no ETFs: 500–600 bonds sourced through partner institutions, of which 100–150 AAA-to-A papers worth around ₹4,000 crore are displayed daily, with the caveat that a credit rating is only one parameter of risk. 09No lock-in, 8.5%, and the liquidity catch 18:42 Correcting what 99% of first-timers get wrong (bonds have no lock-in, pay interest every year and can be sold in the secondary market) then the yield ladder against a 5.5% FD, and the catch that thin liquidity means GoldenPi has to arrange an institution to buy you out. 10Understand the market, then write the code 23:36 The CTO's rule, that you need the headline in your head before you write a line, applied to an OTC market full of variances and shifting rules, which becomes a complex rule-based system so that thousands of retail tickets don't need armies of people. 11Hiring for a problem nobody digitised 27:00 Twenty to twenty-five of about 65 people are in tech, and the pitch to candidates is the rarity of the problem: an age-old industry technology never penetrated, where the hard part is translating market complexity into product, not writing code. 12Node, Python and a second engine 29:50 A best-fit stack chosen for scalability rather than ideology, offered to institutions as plug-and-play SaaS rather than raw APIs, a B2B line that arrived as a market request and is already live with over five institutions, with 12–14 expected by year end. 13What almost shut them down 32:25 Companies die because they quit early or refuse to pivot; GoldenPi's near-death was being alone in a market with nobody else in it, and the early years took a personal toll before timing, the ecosystem and falling FD appeal turned in their favour. 14Who actually buys at ₹10,000 35:03 The core cohort is 30-plus building a fixed-income sleeve, with 24-to-30-year-olds now arriving at the ₹10,000 ticket, and the advice to crypto-curious first-timers is diversification, never putting into one basket what you cannot afford to lose. 15The $100 trillion market nobody watches 38:34 Oil, the Russia-Ukraine war and sovereign borrowing run through bonds: India's government securities market is around $1.2 trillion, the US market $30–40 trillion, global bonds roughly $100 trillion against $65–70 trillion of equity, and with prices falling as yields spike, first-timers get a cheaper entry. 16Shoe Dog, coding, and the meaning of Pi 41:50 Closing on inspirations: a CTO who still codes daily and says stay current or go stale, an autobiography habit that landed on Shoe Dog for its unglamorised downs, and the naming debate that fused gold's Indian association with safety and Pi's multiplication.

Timeline

The arc, briefly

Two engineers taking a phone-call market down to a ten thousand rupee ticket.

2005-06Bonds were for people with a lot of money. Vishal was covering the market around then, when a lot ran ₹10 lakh and up and retail had no way in at all.

2006-07A first attempt at a business, and it fails. Abhijit and Samar had met in college through a common friend, then found themselves on the same project at Ixia Technologies logging 14-hour days on test infrastructure for routers and switches. Their first go at building something together failed for seven reasons, which he counts as a good start because it proved the two of them converge on the same decision.

2016Same tech park, same lane, same day out. Abhijit was at a fintech company in Bangalore when Samar switched jobs into an office in the same tech park and took a house a couple of blocks down the road. They started talking again, decided they had had enough of the corporate world, and resigned on the same day.

2016-17Flip one: the exchange has no bond market. As techies new to the field, the convenient route looked like routing retail bond buying through the exchange, and it turned out the bond market does not live there. What they were solving for was set on day one; only the mechanism kept moving.

2016-17Flip two: a good algorithm and no inventory. The next build assumed a ₹1 lakh ticket and demoed perfectly until it met real market data, because sellers dealt in minimums of ₹50 lakh and multiples of that. So the two went to the financial institutions with an arithmetic argument: aggregated retail would come to more than what the corporate desk brings in. A note to Nithin Kamath's team asking for integration APIs turned into an office visit, a long argument about fixed income, and his backing.

TodayTen thousand rupees buys a bond. Between 500 and 600 papers are sourced through partner institutions, of which 100 to 150 rated AAA to A and worth about ₹4,000 crore go on display every day. Roughly 65 people run it, 20 to 25 of them in tech, and the SaaS line the institutions themselves asked for is live at more than five of them, with 12 to 14 expected by year end.

Takeaways

Ideas to carry out of this hour

The barrier was lot size, not retail appetite

India's bond market kept retail out by arithmetic, not by rule: the market lot ran ₹10 lakh to ₹50 lakh, and sellers dealt in minimums of ₹50 lakh and multiples of it, literally crores. GoldenPi's second iteration built a portfolio algorithm around a ₹1 lakh ticket and worked perfectly until it met real data, because the inventory didn't exist at that size. The company only became possible when financial institutions accepted the argument that aggregated retail participation would be worth more than what corporate desks bring.

▶ 7:29 ▶ 8:49 ▶ 9:37

Retail scale is a technology problem, not a sales problem

As an institution you might do five trades in a day, each ₹50–100 crore, discovered and negotiated over phone calls, a massive market that runs purely on operations. Open that to retail and you are suddenly processing thousands, perhaps lakhs, of small-ticket transactions a day, which no voice-driven desk can absorb. That gap is what GoldenPi built into: a complex rule-based system, explicitly modelled on what e-commerce did to retail, that carries the market's variances so that scaling doesn't mean hiring manpower.

▶ 15:16 ▶ 15:31 ▶ 16:04

A bond is a loan with a coupon, not an FD with a lock-in

The founders say 99% of first-time bond investors get this wrong. There is no lock-in: you can sell into the secondary market whenever you need to, and the issuer pays interest every year until maturity rather than a lump sum at exit. The honest caveat comes immediately after: secondary liquidity in Indian bonds is thin, you need a willing buyer at your price, so GoldenPi typically arranges a financial institution to take the paper off a retail seller.

▶ 18:56 ▶ 19:29 ▶ 21:47

Roughly three points a year, against your fixed deposit

At the time of recording the best FD rates were 5.5–6%, a ten-year central government bond was yielding about 7.3%, and AAA-rated corporate bonds started at 7.8% and ran up to 8.5%. On a four-year holding, the popular FD duration, that three-point gap is put at roughly 12% more on the money. Go down the ladder and AA paper pays 9–10.5% and A-rated around 11%, which is exactly where the founders warn that the rating is only one parameter and the quality of the company matters too.

▶ 20:06 ▶ 20:31 ▶ 21:02

The first year iterates the route, never the problem

The founders are emphatic that the thing they were solving for was set on day one and never moved; what changed three or four times was how. The first flip assumed the exchange was the convenient route and collapsed when they discovered the bond market doesn't exist there. The second assumed a ₹1 lakh ticket and collapsed on inventory. Their advice to founders is to expect that first year to be spent finding the mechanism, not questioning the mission.

▶ 6:15 ▶ 6:44 ▶ 7:45

Institutions want an interface, not your API

The B2B line wasn't planned; it arrived as a market requirement while the consumer business was scaling, and it now runs as SaaS for bond suppliers who are long-established institutions rather than technology shops. Hand them an API and they get nervous, so GoldenPi ships a user-friendly interface where they upload inventory and the system handles everything else. To drive a car you don't need to go inside the engine. Over five institutions were live at the time of recording, with 12–14 expected by year end.

▶ 31:06 ▶ 31:38 ▶ 31:55

The market that sets the world's price is the one retail ignores

Global bonds are roughly a $100 trillion market against $65–70 trillion of global equity, the US alone somewhere in the $30–40 trillion range, and India's government securities market about $1.2 trillion. Sovereign borrowing is how governments fund themselves, which is why oil prices, the Russia-Ukraine war and political instability transmit through bonds first. The investor takeaway is mechanical: prices and yields move inversely, so a period of spiking yields and falling prices is a cheaper entry, not a warning.

▶ 39:53 ▶ 40:23 ▶ 40:52

Being first in a market means there is no market to ride

Asked what nearly ended the company, the answer isn't a funding round or a competitor. It is that when they started there was nobody else in the bond market for retail, so the early scaling took a tremendous personal effort simply to keep the company afloat. Their diagnosis of why startups die is the mirror image: they don't try for long enough, and they won't pivot the business model. What eventually turned was a combination: timing, Nithin Kamath's support, an ecosystem where leaving money in an FD stopped looking obvious.

▶ 33:48 ▶ 34:04 ▶ 35:03

The numbers, drawn

What the episode measures

Every figure below was said on air, with timestamps included and caveats kept.

What the same money earns, by instrument

% per year

Rates as stated in conversation at the time of recording: FDs 5.5–6% (upper bound shown), AAA 7.8–8.5% (upper shown), AA 9–10.5% (midpoint shown), A 'around 11%'. Higher yields carry lower credit quality, the founders' own caveat.▶ 20:31

Debt is the bigger market

$ trillion

Sizes as stated on air: global bonds 'close to 100 trillion', global equity '65 to 70 trillion' (upper shown), the US market described loosely as '30 to 40 trillion' including treasury bills (upper shown), India's G-secs 'close to 1.2 trillion'. Approximate figures quoted from memory in conversation.▶ 40:23

The problem stayed; the route changed

three attempts at the same question

What the founders mean by three or four flips of the model. The thing they were solving for was set on day one and never moved; their advice is to expect the first year to go on finding the mechanism, not on questioning the mission.▶ 7:45

Five phone calls, or a lakh of tickets

what a day looks like on either side of the change

Abhijit calls the bond market a massive business run purely on operations, with prices discovered and negotiated by voice. The dots are drawn to suggest the volume, not to count it. What stands in for the phone call is a rule-based system built so that scaling does not mean hiring people.▶ 15:31

Worth keeping

Lines that stay

The bond market has been a hundreds-of-years-old market, and it has been an institutional game — investors talking over phone calls to find out what bonds are there, negotiating on the prices, getting the delivery done. It's a massive market purely driven on operations.

— Abhijit ▶ 14:46

The algorithm is fine, but you will not find that right inventory to source it.

— Samar ▶ 8:16

We went to the financial institutions and said, would you give us a smaller ticket size? We showed them how retail participation, when aggregated, would be way more than what corporate does.

— Abhijit ▶ 9:37

Even before you write code you need to understand what you are writing for. It's like you need to have a headline in your head — then you start writing.

— Samar ▶ 24:11

To drive a car you don't need to go inside the engine. We have made it a SaaS way — the entire complexity is behind the scene.

— Samar ▶ 31:06

Clips that travel

Short on time? Start here

First-time investors who assume bonds are for institutions

The lot size that locked retail out

The pivot that defined the company: an algorithm with no inventory behind it, and the pitch that aggregated retail beats the corporate desk.

7:29 → 10:07 · 3 min ▶ Watch clip

Equity investors who have never bought a bond

Ownership versus a loan, in under two minutes

The cleanest explainer in the episode of why a bondholder's risk is structurally different from a shareholder's.

10:07 → 11:56 · 2 min ▶ Watch clip

Founders digitising an analogue industry

A phone-call market meets retail scale

Five trades of ₹50–100 crore a day against lakhs of retail tickets, the exact sentence where the technology thesis becomes obvious.

14:46 → 16:20 · 2 min ▶ Watch clip

Savers whose fixed income is entirely in FDs

8.5% against a 5.5% fixed deposit

The full yield ladder from G-secs to A-rated paper, plus the liquidity caveat nobody tells first-timers.

19:58 → 22:32 · 3 min ▶ Watch clip

Investors reading macro headlines

The $100 trillion market and the yield spike

Why oil and war transmit through bonds, how big the market actually is, and why falling prices are an entry rather than an alarm.

38:34 → 41:50 · 3 min ▶ Watch clip

Glossary

The jargon, unpacked

Bond

A tradeable loan to a company or government: you hand over capital, the issuer is contractually obliged to pay a fixed rate of interest each year and return the principal at maturity.

Fixed income

The asset class of instruments paying a predetermined return (bonds, debentures, government securities) held as the portion of a portfolio that shouldn't move with equity markets.

Credit rating

The AAA-to-D scale that grades an issuer's likelihood of paying you back; GoldenPi lists only AAA to A, and the founders insist it is one parameter of risk, not the whole picture.

Market lot

The minimum quantity a seller will trade in. In Indian bonds it ran ₹10 lakh to ₹50 lakh and multiples, the single number that kept retail investors out.

OTC market

Over-the-counter: a market with no central exchange, where buyers and sellers find each other and negotiate bilaterally, historically by phone, which is why the bond market never digitised.

Secondary market

Where an existing bond is resold before maturity. There is no lock-in, but Indian bond liquidity is thin, so a platform may have to source an institutional buyer for a retail seller.

G-Sec

Government securities, meaning sovereign bonds and treasury bills issued to fund government spending. India's G-Sec market is put at about $1.2 trillion in the conversation.

Yield-price inversion

Because a bond's coupon is fixed, paying less for the paper means earning a higher return, so when prices fall, yields rise, and vice versa.

Connections

If this resonated, go here next

EP 49 · Fintech

Full transcript

The whole conversation, searchable

Auto-generated captions, lightly cleaned. Click a timestamp to open that moment on YouTube.