Now: Pre-Seed Investor @DeVC_Global || Prev: Founder @VerakInsurance (acq. by ID) || Views are my own

BOM šŸ” BLR šŸ¢SFO šŸ›« NCR šŸ­
I’ve automated 80% of my 2024 desk job. AI now saves me 600+ hrs per yr. On 30th June 2025, I shared my 10,000 hours of AI goal. I’m now 750+ hours into this journey. My set-up costs ~$12k per annum and saves ~$60k in notional labor costs i.e. a base case ROI of 5x on labor cost savings and upside benefit from faster & 24x7 work completion. I now have ~3.5 hrs of additional time per day to focus on outbound activities e.g. getting on the plane, meeting founders & experts F2F, going to events etc Here’s a quick summary of my work OS:
13
17
418
57,467
Rahul Mathur retweeted
DeVC and @Rahul_J_Mathur are a beast. We’re probably one of their smallest and earlier cheques, but the leverage they give early tech founders is crazy... Unlocked massive AI credits, instant intros whenever we need them, and zero performative nonsense. Super. High. ROI. pre seed VC. A ton of the technical capability that we've built internally for Kreo would not have been possible without them
2
3
72
6,122
Rahul Mathur retweeted
I spent the last 6 years building a sustainable advisory practice. My team. My service delivery. My claim support at 11pm. All of it was built on a commission structure that the IRDAI proposal just made unviable. (1/3) #IRDAI #Insurance
2
2
7
1,451
Rahul Mathur retweeted
The original Shaadi.com was a side project launched by 23 year old Manish Jain. He paid $100 on 30th July 1996 to buy the domain. He built V1 of the website in Microsoft FrontPage during nights & weekends. His day job was working at Andersen Consulting (now called Accenture) as a fresh college graduate from Indiana University The website was launched (officially) on 1st October 1996. Manish used email marketing (before spam became a thing) to solicit NRIs to create profiles. In less than a year, Shaadi was adding 75 new profiles per week with 1,500 weekly visitors. There were no paid profiles. Manish was pitching $99 per month to run banner ads on his website. This is long before Anupam Mittal came into the picture ā¤µļø
12
21
275
41,684
The original Shaadi.com was a side project launched by 23 year old Manish Jain. He paid $100 on 30th July 1996 to buy the domain. He built V1 of the website in Microsoft FrontPage during nights & weekends. His day job was working at Andersen Consulting (now called Accenture) as a fresh college graduate from Indiana University The website was launched (officially) on 1st October 1996. Manish used email marketing (before spam became a thing) to solicit NRIs to create profiles. In less than a year, Shaadi was adding 75 new profiles per week with 1,500 weekly visitors. There were no paid profiles. Manish was pitching $99 per month to run banner ads on his website. This is long before Anupam Mittal came into the picture ā¤µļø
12
21
275
41,684
Fast forward to 2000, Manish had switched to Cisco. He was approached by Siddharth Mehta to sell the domain. Manish agreed to sell this domain. Siddharth flew to Los Angeles to execute the deal. Now, this is when Anupam Mittal comes into the picture. Anupam had started Sagaai.com in 1997. Anupam offered $5,000 to Siddharth for the domain. He declined & the bid kept going up over the next 2 months In 2001, People Interactive (Anupam’s company & now owner of Shaadi) agreed to pay $25,000 which is literally 83% of the total funds that the company had 🤯 I used the Wayback Machine to find Manish’s coded website as of 14th April 1997 (which is 8 months before I was born!)
1
32
5,075
The story behind this domain has both sellers remorse (Manish said ā€œI sold a phenomenal domain for a pittanceā€) and buyers terror (Anupam put 80% of cash into the domain) Manish built V1 of this iconic website in 1996 when Hotmail was just launched & India had 4.5 Lakh internet users. By the time he sold the website in 2000, our internet population grew to 56 Lakh users (0.5% penetration) By the time Wharton profiled Anupam Mittal, India’s internet population grew to 14.2 crore (11% penetration). Anupam raised his first VC round 10 yrs after starting the company. Anupam was early. Manish was even earlier. Manish had learnt his lesson: He owns several such single-world category domain names to this date. When I met him at GFF 2026, he told me that someone offered him $500k for one of these domains, which he flat out declined (apparently the bidder started at $5k šŸ˜†) We’re very lucky that Manish has catalogued this entire journey on his blog & made the original marketing collateral for the website available (this is 30+ yrs old now): celestri.org/2026/07/30/shaa…
1
2
66
4,363
Very proud of my flat mate NS! He's spent the past 6 months deep in the US Tax space. Broke the benchmark as a solo dev earlier last month & recently went viral for finding an extremely smart use of Jev in Tax It has been incredible to see his journey over the past 5 years of knowing him. Do give him a follow & if you have any US tax queries, he is your go to person šŸ˜‡
We've been building Kyoto, an AI for your taxes. It scores 100% on TaxCalcBench v2: all 50 test cases correct across federal and state returns, on every scored line. The best frontier model scores 64%.
2
4
130
20,236
I'm down 5% portfolio level today driven by PB FinTech Concentration has some evils; it is painful to see my 2nd largest position drop 30% (so far) today with (what I believe) will be another 4% fall later today. Effectively, 3 yrs of averaging in at a great price got wiped clean. I've also read through the 2-part IRDAI note. I'm sure they'll hear a good bit from industry participants. Felt more like retaliation rather than regulation (latter being the typical stance) Some leaders are yet to realize that Distribution is not cheap. Intermediation isn't disappearing overnight because of a Govt website or AI ONDC was a great lesson in this regard. FWIW, I'll average down across today, tomorrow & Monday. I believe calmer & more sensible heads will ultimately prevail šŸ˜‡
This is what a black swan like event looks like Policybazaar & Turtlemint are both down 20% today after IRDAI’s proposed changes to insurance distribution economics One consultation paper comes out & the market can reprice the entire business model within minutes
36
7
159
65,626
Of course, don't treat this as financial advice. DYOR. If you want a fun statistic: this position in PB was built across 32 independent buys over 800+ trading days. XIRR was close to 25% until this morning šŸ˜‚
2
10
2,631
I am delighted to see our Govt walk the talk on supporting DeepTech in India Last week, I was going through the literature of the ₹10,000 crore SIDBI FoF 2.0 scheme and I noticed a few positive changes. DeepTech is hard (both for the founder & the VC). Underwriting these businesses during today’s boom era requires even more discipline, domain depth & patience. Through the FoF 2.0 scheme, the Govt of India (through SIDBI) provides capital to VC funds to invest downstream into DPIIT registered startups. The new scheme has 3 key provisions to help DeepTech VCs:
9
41
309
15,405
These provisions are as follows: (1) Max fund life has been extended from 12 to 18 years for DeepTech AIFs. Critical for gestation heavy sectors like Semiconductors (e.g. DeVC portfolio co Morphing Machines was founded in Nov 2005 and did their 1st tape out this month!) (2) The investment multiplier (i.e. how much non-Govt capital must be invested in the fund) is reduced from 2x to 1.5x only for DeepTech AIFs. Raising funds is hard, this reduces the barrier which is very helpful for emerging fund managers (which is most Deeptech VCs) (3) The max contribution from Govt to DeepTech AIFs has been doubled from ~20% of corpus to 40% of corpus only for DeepTech AIFs! Patient capital behind patient investors šŸ˜‡ This will be a major catalyst for DeepTech in India
2
2
22
1,505
Through the FFS 1.0 scheme, Govt of India allocated ₹800 cr across ~10 DeepTech VC funds. In total, all VC funds backed by FFS 1.0 invested ~₹4,900 cr into 280 DeepTech startups (according to ISCI in May 2026) Therefore, startups backed by FFS 1.0 include Agnikul Cosmos, Digantara, Pixxel, Dhruva Space, EtherealX, Sarla Aviation & several others. When FFS 1.0 was conceived in 2016, DeepTech in India was a mirage, we didn’t even have a space policy. As FoF 2.0 rolls out in 2026, DeepTech in India is real: Our 1st unicorn (Skyroot), our 1st IPO (SEDEMAC), our 1st ₹1000+ cr domestic order (EO PPP) & our 1st export orders. A nation compounds over decades. We won’t see the results of FoF 2.0 for at least another decade.
1
3
25
1,083
How do you retain cable TV subscribers when they are spending more time watching Reels? Hathway’s answer is to put the Reels on cable TV šŸ˜‚ Hathway’s default TV channel is now a Reels feed. I discovered this earlier last week when I switched the TV on in my grandfather’s home after many months. Reels Magic was launched on 25th March 2025. You can submit a 30-90 second Reel of yourself via WhatsApp or email to appear on the TV channel. If you have a Hathway connection, you’re effectively forced to watch the channel for ~30 seconds until you can switch to the actual channel you wish to watch. This channel is owned by a listed company called DEN Networks. RIL took a 66% stake in the company for ~₹2,700 crore around the same time it bought Hathway (therefore, part of the same RIL umbrella entity) The concept is rock solid ā¤µļø
5
55
5,283
Reels Magic is fundamentally UGC being broadcasted on cable TV: (1) The idea isn’t new: Comcast in US had launched Current TV back in August 2005 to allow its audience to send ā€œpodsā€ (short clips about news, drama, romance etc) with a chance to feature on the channel. (2) The channel plays to a Status game: Having your Reel featured on TV is more high status than getting 1M views (although, the latter might help you get a brand deal or a real job šŸ˜†). (3) These UGC channels can be money machines: user submitted content (UGC) means free cost of production (unlike other channels which pay to license content). These channels have the monetization via ads. So, lower COGS & distribution cost (some virality) for similar ad rack rates! However, Hathway is trying to solve a 2026 era engagement problem with their 2000s era digital asset.
1
5
625
The traction isn’t as great as the underlying concept. I’ve broadly seen the same set of Reels over a few days (a clear indication that not enough people are submitting new content). The fundamental issue is that DTH & cable TV subscriber count is dropping across India. Hathway’s active cable subscribers have declined 20% from 56.6 Lakh in March 2021 to 45.1 Lakh in March 2026. While BARC data claims that average time spent on TV hasn’t declined, I think time spent reality would look a lot more grim than subscriber count. Hathway & its peers are fighting a losing battle. Its competition includes algo gods like Netflix & Instagram (the latter has come to TV as well). In my BLR home, we don’t have a DTH or cable TV subscription. In our Mumbai home, my grandfather retains it for posterity (or more likely because he doesn’t know how to cancel the subscription).
5
455
Rahul Mathur retweeted
The obvious is missing. So we built Sol - hellosol.app Sol finds the work itself, does it, and comes back for your approval. Every day in our emails we say "I’ll shareā€, "I'll reviewā€, "I'll get back" - then repeat the exact same thing to an AI. Why? Sol finds everything you said you’d do & gets them started for you. It does the research, creates the doc, builds the slides, finds the time, connects the dots across multiple emails, doing everything it takes to get the job done - but doesn’t send, schedule, or share anything until you approve. Sol runs on its own computer, uses a browser, and has a library of skills that automatically get assigned to the work that needs to get done. No setup. It just starts working. We've raised $4M from General Catalyst, Nexus Venture Partners, DeVC, PeerCheque, Kunal Shah, and a few others. Extending early access now. @generalcatalyst @nexusvp @DeVC_Global @peercheque @neerajarora @b_jishnu @kunalb11 @miten @RTinkslinger @Rahul_J_Mathur @AkarshS27 @SiddhantD06 @RajatAgarwal167
177
74
540
651,255
MDR fees attract 18% GST Here’s my best attempt at explaining this concept using a simple UPI transaction I pay ₹10,000/- via UPI QR to buy the New Balance 574 from their store on 100ft road in Indirangar Effective 15th Oct ā€˜26, an MDR of 0.4% (capped at ₹300) is charged on UPI payment to a merchant where the value is > ₹2k (with some more caveats) MDR = 0.4% * ₹10k = ₹40 GST on MDR = 18% * ₹40 = ₹7.20 New Balance would now collect ₹9,952.80 from my payment Therefore, the ā€œrealā€ MDR rate is 0.472%. It means nothing for a big brand with high margins like New Balance. But, this poses a nightmare for smaller merchants ā¤µļø
41
36
484
94,558
Here is the problem for smaller merchants who are not GST registered: Per my calculations, approx. 17.7% of businesses are GST registered Without GST registration, it is not possible to claim ITC (Input Tax Credit) and therefore, this additional GST on MDR has to be absorbed as a cost of doing business. Merchants who bill less than ₹1L per month via UPI are exempt from MDR For those who are above this threshold, the additional cost on transactions above ₹2k imposed by MDR + GST is ~0.472% Trivial if you run a high gross margin service business but painful if you run a thin gross margin retailing business.
2
37
6,609
In fact, restaurants which pay 5% GST are negatively impacted by this MDR since they cannot claim the ITC which New Balance & other companies paying 18% GST can. Therefore, cloud kitchens like The Mumbai Pav co (which my mom has invested in) would be adversely impacted by the GST charged on MDR The challenge with imposing MDR on UPI runs much deeper than this GST edge case which I shared. Fundamentally, it is extremely difficult to charge for a previously ā€œfreeā€ service (due to a concept called the Prince Anchoring). There is always outrage when something goes from ā€œfreeā€ to even a nominal charge. The public backlash is expected but the industry desperately needs the MDR to make their payments business survive without Govt incentive programs. My grandfather summarized this beautifully last evening: ā€œeither way, the taxpayer pays, which has been true since I started paying taxes in 1958ā€ šŸ˜†
7
1
54
5,388