Z47 Moments

Srikanth Velamakanni built an AI company before most people knew the A in AI.

In 2000, long before ChatGPT, he and five co-founders started Fractal Analytics with a simple bet: that math could predict human behaviour and help companies make better decisions.

The market wasn't ready. HDFC took 18 months to convince. Citibank made them prove themselves against 100 in-house PhDs. They won on accuracy, stayed profitable, and 26 years later Fractal became India's first AI company to go public.

In this episode of Unstarted, Srikanth sits down with Avnish to talk about the story behind the story. Growing up middle class in a small oil town in Assam. Why he never thought he was founder material, and the Narayana Murthy lecture that changed his mind.

If you are building, or still unstarted, this one is full of hard-won lessons on conviction, timing, co-founders, and staying relevant across 26 years of technological change.

Chapers

0:00 Intro
1:15 From a small town in Assam to Fractal
2:47 Math plus human behaviour: an AI company before AI
3:34 "I never thought I would be an entrepreneur"
5:01 The Narayana Murthy moment
5:54 Not money: building a 100 year company
6:57 Conviction vs market readiness
9:32 When the market is not ready (ICICI, HDFC)
10:27 The Citibank bake-off against 100 PhDs
12:00 Compounding your core passion
15:44 The two co-founder types you actually need
18:39 Finding your ICP without chasing noise
19:16 26 years of AI, from regression to transformers
23:08 Strategy is what you refuse to do
26:24 Elected CEO, then asked to leave in 6 months
28:02 Defining the company's values after crisis
29:19 Raising money and investor expectations
31:34 Will AI make Fractal obsolete?
33:58 Why Fractal was built in Mumbai
36:30 Advice: build what OpenAI cannot kill

Amit Lakhotia moved from a village in Churu, Rajasthan to Delhi at 10, speaking Marwari and barely any Hindi. Twenty people shared three rooms. His father couldn't afford the school fees, so he made a bet with the principal: if his son didn't come first within a year, throw him out.

That year Amit scored 19 out of 20 in Maths and Hindi, and 1 out of 20 in English. His tuition was waived from class four through twelve.

He went on to NSIT and IIM Ahmedabad, helped scale MakeMyTrip, Paytm and Tokopedia, built a taxi business that was four times Ola's size before shutting it down, and then founded Park+ — which raised its first cheque two months before COVID shut down every mall and office in the country.

In this episode of Unstarted, Amit talks to Avnish Bajaj about when to walk away from a working business, the vitamin versus painkiller test for choosing what to build, why he picked parking as the wedge into the car ownership market, how access control emerged from the worst possible timing, and why brand and revenue aren't a trade-off if you think creatively.

Unstarted is a series for founders, by founders, with questions crowdsourced from the audience.

Guest: Amit Lakhotia, Founder & CEO, Park+
Host: Avnish Bajaj, Founder & Managing Director, Z47

Chapters

0:00 Intro: the founder who gets away with scolding me
1:14 Churu to Delhi, and 20 people in three rooms
2:09 Arriving at school with no English
3:04 The bet his father made with the principal
5:01 How he survived that first year
5:47 19 out of 20 in Maths. 1 out of 20 in English.
7:21 NSIT, IIM-A, and where the confidence came from
9:20 Sitting in his father's shop and deciding to drop out
10:12 Turning around the IIM-A souvenir store
12:15 GetMeCab: profitable in month three
13:42 Why he walked away from a business 4x Ola's size
14:50 Q: I want to start a business. How?
17:57 Vitamins vs painkillers
18:38 The two ideas he shortlisted, and why parking won
21:24 Q: How do you measure progress during a pivot?
21:36 Two months in, COVID arrives
22:52 Turning a pandemic into the access control business
24:38 The metric that mattered: client acceptance
25:35 Revenue vs brand, and getting signage for free
27:43 Closing: build the network, and give before you take

Nikita Khanna spent fifteen years straightening her hair before she realised the problem wasn't her hair. It was that nobody had built products for it.

She was on partner track at McKinsey her only job ever when she started going down YouTube rabbit holes during COVID, trying to find a sustainable way to manage wavy, frizz-prone hair. She found the answers in Brazil: similar hair texture, dozens of local brands, none of it available in India. When she asked people in the industry whether it was a business, almost everyone told her the category was too small and too random.

Roughly 80% of Indians have hair somewhere on the wavy-to-curly spectrum. She built Moxie anyway.

In this episode of Unstarted, Nikita talks to Avnish Bajaj about validating an idea before you quit your job, the ABC framework she now uses to evaluate any new category, why she never thought she was cut out to be a founder, learning to delegate and build an org for the first time, running a company with her husband as co-founder, and what it actually takes to build a brand in a market where launching a product has never been easier.

Unstarted is a series for founders, by founders, with questions crowdsourced from the audience.

Chapters 

0:00 Intro: the company we should have invested in
00:00 Growing up in Delhi, watching her father build a business
00:00 DU, XLRI, and landing at McKinsey
00:00 Straightening her hair since 13, and deciding to stop
00:00 Finding the answer in Brazil
00:00 "Am I even cut out to be a founder?"
00:00 Q: How do I validate an idea before I quit?
00:00 The ABC framework: assortment, business, consumer
00:00 Everyone said the category was too small
00:00 Q: How does a young founder become a CEO?
00:00 Building with her husband as co-founder
00:00 Why viral products aren't the goal

Aman Goel, founder of Greylabs.ai, joins Avnish on Unstarted for one of the most relatable founder journeys yet — a middle-class kid from Kanpur who was hustling for money from his first year of college and never stopped.

Aman is refreshingly honest that his motivation was simply to make money. He earned his first ₹4,000 solving JEE papers, built a content-writing business with his now-wife and co-founder Harshita, and had saved close to ₹50 lakh before he even graduated. 

He went on to build Cogno AI - an AI chatbot and voice company for banks - in 2017, before generative AI existed, bootstrapped it to a million-dollar revenue run rate, and sold it to Exotel at 25.

Now with Greylabs.ai, he's building voice AI for India's banks and financial institutions. 

In this episode he breaks down finding customers in an industry you have no network in, the "student not salesperson" playbook that got him into SBI and HDFC, why he and Harshita split the company cleanly into "she builds, I sell," how to part ways with a co-founder without a fight, picking a razor-sharp ICP, and why in a world of weekly AI launches trust is still the only thing that sells.

Chapters

00:00 Intro — a genuinely relatable founder story
01:00 Growing up middle-class in Kanpur
02:50 "My motivation was money" — where ambition came from
04:00 The internet, Silicon Valley, and deciding not to settle in the US
05:50 Earning in college: ₹4,000 to ₹50 lakh before graduating
08:20 How to create your own opportunity in college
11:50 Starting Cogno AI — and the RED vs REL framework
15:40 Building AI in 2017, before ChatGPT
17:20 Getting into banks: the "student, not salesperson" playbook
21:40 Negotiating with SBI as two 20-year-olds
24:40 The first customer, and the referrals that followed
25:30 The co-founder exit: quick decision, generous separation
27:20 Harshita joins: "she builds, I sell"
30:40 The exit to Exotel — doubling the ask in 10 minutes
34:40 Finding the market first: why he stayed in India
37:50 Founder-market fit and picking a razor-sharp ICP
40:40 Can AI disrupt his AI business? Why trust wins
42:50 Eating your own dog food: building internal tools on Replit
44:40 AI adoption in banks — from pilots to 50 lakh calls a month
46:14 The coming pricing pressure, and how service is the moat
49:20 Closing advice: persistence is the one thing that can't be replicated

Sameer Nigam, founder and CEO of PhonePe, joins Unstarted for a candid conversation on building a company that now serves over 700 million users.

From a middle-class naval family to a failed music startup (Mind360), an acquisition by Flipkart, and the founding of PhonePe, Sameer breaks down the thinking behind his "asymmetric bets," why he chose UPI when the entire market was chasing wallets, and how the Big Billion Day payments collapse revealed the problem he'd spend the next decade solving.

He also opens up on his 31-year partnership with co-founder Rahul Chari, why he embraces regulators instead of fighting them, what PhonePe's data reveals about how India really spends, and the story behind the IPO that was pulled at the last minute.

A must-watch for founders thinking about co-founders, market timing, building for population scale, and knowing when to cut your losses.

Unstarted is a show about founders, for founders hosted by Avnish Bajaj. 

Chapters

0:00 Intro: meet Sameer Nigam & the "celebrity founder" tag
2:15 Family background: a Navy dad and an entrepreneur mom at early TCS
10:45 Why tech fundamentals still win (and the Google "75% AI code" point)
17:40 First venture: Mind360, the "iTunes for India"
22:10 Why it failed: wrong market sizing and the piracy era
26:00 Acquired by Flipkart and the B2B pivot
34:15 Big Billion Day: when India's payment infra collapsed
38:00 Why wallets were never the answer
41:20 The asymmetric bet on UPI
44:30 Ghar Wapsi: merging back into Flipkart to buy time
47:10 Demonetization: luck, timing, and near-zero competition
50:00 Obsessing over infrastructure and the right foundation
52:30 ICP, population scale, and the "painkiller in the hinterland"
55:40 Incumbents push back: ICICI blocking and the RBI visit
59:00 What PhonePe's data reveals about how India spends
1:05:00 The most anticipated IPO that didn't happen
1:08:30 Final advice: don't be an entrepreneur by FOMO

Saahil Goel, co-founder and CEO of Shiprocket, joins us for a candid deep-dive into how a self-taught coder from Delhi built the logistics and enablement backbone powering a huge chunk of India's direct-to-consumer commerce.

Sahil takes us all the way back selling his dot matrix printer on Baazee as a kid, teaching himself HTML by hacking "view source," and building websites for cash in college.

Along the way he shares the hard-won lessons behind every pivot: why Indian SMEs won't pay upfront, why you should never fall in love with your product, and how obsession with the merchant's problem kept the company alive through years when every investor said the market didn't exist.

We also get into co-founder dynamics and trust, what actually changed scaling from 25 to 1,500 people, why culture is the "uncodified personality" of a company set by the founder's actions, and being a "microsurgeon, not a micromanager." 

Finally, Saahil lays out his conviction on AI - from writing 70%+ of code with AI, to Shiprocket's in-house voice model built for India's 22 languages, to the "AI for Bharat" mission of reaching the next 58 million merchants.

A rich, practical episode for anyone still unstarted.

What they talk about:

1. From selling a printer on Baazee and self-taught coding to building Shiprocket into India's commerce shipping backbone

2. The real story behind the pivots  KartRocket to Kraftly to Shiprocket and why you stay stubborn with the problem, not the product

3. What PMF actually feels like, plus co-founders, trust, and scaling culture from 25 to 1,500 people

4. His bet on AI: writing most of the code with AI, an in-house voice model for India's 22 languages, and the mission to reach 60M MSMEs

00:00 Intro - Saahil Goel and Shiprocket
02:30 Delhi boy, self-taught coder & selling a printer on Baazee
07:00 Business family & knowing early he'd be a founder
12:00 Meeting Gautam & the start of Bigfoot Retail Solutions
17:00 Co-founders: trust, complementary skills & full commitment
23:00 PMF is a journey, not a destination
29:00 Stubborn with the problem, not the product
35:00 CartRocket to Craftly: the pivots and hard lessons
42:00 Building Shiprocket & why B2B behaved like consumer
48:00 Scaling 25 to 1,500: culture, micro-surgery & builder mindset
55:00 Bias to action vs bias for excellence
60:00 AI at Shiprocket: 70% of code, voice for Bharat & 60M MSMEs
70:00 The IPO ahead & advice for the still-unstarted

Our first-ever episode with co-founders. Karan and Himanshu of WootzWork join Avnish to talk about building a hardware and manufacturing exports company out of India — and why they leaned into a market everyone told them wouldn't work.

From an investment banking and Stanford GSB background (Karan) and IIT Kharagpur mechanical engineering (Himanshu), the two met at a previous company where one interviewed the other as head of engineering a one-hour conversation that ran three hours. In this episode they unpack how a single insult ("we don't talk to Indian engineers, we work with Germans") lit the fire, how they closed their first order from India with no factory of their own, and why the naysayers became their strongest signal to go ahead.

What you'll learn in this episode:

1. Why starting in a market everyone doubts can be an advantage — and how to tell real naysaying from noise
2. How to win your first customer with no factory, no brand, and no track record
3. The co-founder decision that actually matters: why the answer is never "the middle"
4. How to choose a co-founder — honesty and complementary skills over everything else
5. Why customer obsession beats worrying about network or capital early on
6. Manufacturing as a system: how fragmentation held India back and how to fix it
7. Scaling global sales by hiring domain experts in their 40s–50s who are local to the market

Chapters

00:00 Intro 
02:00 Karan's path: banking, AirAsia India, Stanford GSB
04:30 Himanshu's path: IIT Kharagpur, ITC, and the machines
05:15 "We don't talk to Indian engineers" — the moment that lit the fire
06:30 The China vs India buying experience: what's actually broken
08:00 Why the market was ripe — and betting on exports
11:45 Customer obsession: getting the first customer in
13:15 The co-founder dynamic: why the answer is never in the middle
20:25 Turning the naysaying into conviction
22:30 The first order: won with no factory
23:12 Why the customer still needs them: engineering vs manufacturing
28:30 Managing people with 40–50 years of experience
30:30 AI and robotics in the manufacturing story
33:40 Closing: why 2.3 is the right number of co-founders

What happens if you fail with everyone's money?

Abhishek Bansal never let himself ask and he thinks that's the only reason Shadowfax exists.
He's the first public-company CEO on Unstarted: a middle-class kid from Meerut who quit his job without telling his parents, asked a VC for $5 million with zero orders, and was the only one still standing after the year every competitor died.
In this conversation with Avnish Bajaj, they attempt to answer:

1. Should you have a co-founder and how do you actually pick one? 
2. What should founders look for in a VC when there are multiple offers? 3. How do you tell your team you're failing?
4. Sustainable growth, or aggressive scaling?

By the end, you understand his test: accept the failure, name the core problem, tell your team. Because teams don't break on failure — they break on how you behave in it.

Chapters

0:00 Intro: our first public company CEO
1:19 Meerut, a middle-class home, and IIT Delhi
4:04 Parents didn't know he had quit his job
7:50 Q: How do you choose between VC offers?
8:47 Why he wanted a co-founder, not a skills match
11:03 Why not bootstrap?
11:28 Asking for $5 million with no product
12:23 The ₹8 lakh phone call
13:28 The angels who never sold: 1000x and counting
14:31 Three offers: $2M, $5M, $9M
17:19 Q: How do founders bounce back from failure?
17:46 The 2015 crash: five funded, one survivor
23:07 Knowing the core problem is the hardest problem
28:07 The offsite: profitable in three months
28:32 The retention numbers behind the culture
29:20 Q: Aggressive scaling or sustainable growth?
29:56 "In India, no idea is unique"
32:14 The guardrails: unit economics and 50-30-20
37:26 The CAPEX model that ran 100 iterations
39:07 When the HR team built its own hiring tool
40:20 Closing: teams compound, individuals don't

Most people assume serious AI silicon can only be built by a giant, in the cloud, burning hundreds of millions of dollars. Ravi Annavajhala did the opposite. Kinara built two AI chips for under $50M, 90% of it out of Hyderabad, and sold to NXP, one of the world's largest semiconductor companies for $307M,  all cash. It's one of the largest deep tech exits India has seen, and one of the least talked about.

In this episode, Vikram Vaidyanathan sits down with Ravi to take apart how it actually happened, and what it says about where AI is heading next.

You'll hear: 
1. Why the disruption playbook that minted the NAND-flash exits is the same one that explains Kinara's sale to NXP 
2. The clearest plain-language explanation of training vs. inference you'll find, and why inference is moving off the cloud and onto the edge
3. The four reasons edge beats cloud: cost, latency, privacy, reliability 
4. How you actually shrink a frontier model to run on a chip: distillation, compression, purpose-built retraining 
5. Why India's real AI edge isn't talent or capital, but the industrial data sitting in its factories — and why it can't afford to become a "data colony" 
6. Ravi's distinction between frugal and cheap, and why low cost has stopped meaning low capability

▶ If you're building, investing in, or just tracking India's deep tech story, this is the episode to watch.

About Intelligent Indians Intelligent Indians is Z47's thesis podcast on the people building India's technology future, for builders, operators, and investors who want the operator's view, not the press-release version.
🔔 Subscribe for new episodes.

Chapters 

0:00 A $307M deep-tech exit you didn't hear about
0:45 Meet Ravi Annavajhala & the Kinara story
3:10 The NAND flash lesson: how startups cash in on disruption
5:45 2015: betting on edge AI while everyone chased the data center
7:00 CoreViz → Deep Vision → Kinara: the naming journey
9:15 Why 90% was built in Hyderabad — a talent bet, not a cost cut
11:30 Two chips under $50M: engineering for capital efficiency
15:20 What "the edge" actually is: training vs inference
18:00 Why edge beats cloud: latency, cost, privacy & reliability
22:45 Physical AI: robots, self-driving & the real bottlenecks
28:30 Advice to VCs: conviction, local demand & the infra gap
31:20 Frugal vs cheap: why low cost ≠ low capability
33:30 Closing: India on the cusp of the edge-AI wave

How much money is enough to make starting up safe? Vivek Sinha thought he knew: ₹50 lakh in the bank. The day he hit it, he quit.
Vivek is the founder of Emversity, which trains people for the jobs India actually has in healthcare, hospitality, EPC and manufacturing. 

Homeschooled till eight in a Bihar village surrounded by three rivers, he built Buildzar, lost ₹50 lakh, and spent nine years compounding skills at Mobikwik, OYO and Unacademy before starting again in a sector everyone had written off.

In this episode, Avnish and Vivek take on, 

1. Where did you get the confidence to do this?
2. Why didn't it work? What did you think going in that didn't play out?
3. Should I drop out and work full time on my startup?" 
4. How do I find the right co-founder?
5. How can we secure investment in a stigmatized sector despite proven traction?"
6. What stays with you: build it for the right reasons 

Chapters 

00:00 Introduction
02:00 Flood village to NIT: growing up in Bihar
06:00 First startup & why it failed: market timing
11:00 MobiKwik: 0 to 20 lakh merchants in 6 months
16:00 OYO & Unacademy: the lessons that compounded
22:00 The 9-year gap: how he killed bad ideas early
28:00 Should you drop out of college?
34:00 How many co-founders — and going solo
40:00 Building in a stigmatised, out-of-favour sector
48:00 What Mversity does + closing advice