Parag Arora

Loved watching Andrej Karpathy talk at AI Startup School. Must watch for everyone in Product and Engg to understand software in the AI era 1. Software is undergoing a fundamental shift, moving beyond traditional code. 2. LLMs share characteristics with utilities, fabs, and especially operating systems. 3. LLMs possess "human-like psychology" with both superpowers and cognitive deficits. 4. The future of software is in "partial autonomy apps" and human-AI cooperation. 5. This is the decade of AI Agents. Time to build :) https://lnkd.in/gHQ7Ad9a

124

5 Comments

The Indian startup ecosystem is a shared responsibility - and nobody wins or loses in isolation. Two edtechs scale with hundreds of millions in capital and burn down the whole edtech ecosystem and make it exponentially hard for anyone else trying to build meaningfully in the space. Two qcomms scale and IPO and open doors for a host others to get easier access to capital. The success and failures are intertwined like no other place. This is why VCs also must realize that winning is not a zero sum game at all. Hence, as I build First and Fast Capital - I don’t optimize for my outcomes. I optimize for true ecosystem support. If a company is solid but not in my investment thesis, I personally call another GP and make the warm referral. If a portco can’t crack different parts of the puzzle, I aim to enable everything in my network to help bridge that gap. If a company is at a stage I can’t fund but it is something that deserves support, I focus on making the right connections even if that diverts time from my core focus. When another GP has a solid company that’s too “early” for them, I’d be glad for them to flick it my way. If a set of founders are amazing builders but ordinary storytellers, I let them write the story that I can then tell on their behalf. The true unlock in the Indian early stage ecosystem is enabling and mobilizing any and all support with no expectation of a “return favor.” Because when you truly support the ecosystem and the founders at the center of it, outsized outcomes simply follow. There is no other path and there is no better way.

71

12 Comments

I'm a puzzle-loving, first-principles thinker. A terminal-first hacker. And one thing I cannot stand is leaving broken things alone. Yet for all my years building products, one thing kept nagging me: design hands off a mockup, engineering rebuilds it in code, the two drift apart. Ship. Repeat. Every product I've ever built had this friction baked in - but I'd stopped questioning it. That's just how things worked. Then Kushagra Sinha reached out about a design tool he was building. I'd just closed a round for GMetri and was too deep in to pay attention. He came back a few months later. Walked me through the actual problem. I couldn't stop thinking about it. Then he said "just come to the office for a few days as an advisor." I went. Between Aditya Bandi's vision and the team they'd assembled, a small group punching way above its weight, I got that familiar itch. The one where there's a puzzle in front of you and you can feel there's an angle no one's tried yet. Those few days turned into the most fulfilling stretch of building in my career. Today I'm proud to share: I've joined Noon as CTO, and we're coming out of stealth with $44M from Chemistry, First Round Capital, Scribble Ventures, SV Angel, Afore Capital, Elevation Capital, Homebrew Noon is a dual-canvas where you design how a product looks and how it works, directly on top of your production code. What you design is what ships. The gap I'd accepted as permanent? Noon eliminates it. After years of building on both sides of that divide, getting to build the tool that closes it feels like exactly where I should be. Sign up for early access: noon.design

62

13 Comments

𝗛𝗲𝗿𝗲 𝗮𝗿𝗲 𝘀𝗼𝗺𝗲 𝗴𝗼𝗹𝗱𝗲𝗻 𝘄𝗼𝗿𝗱𝘀 𝗳𝗿𝗼𝗺 𝗺𝘆 𝘃𝗶𝗿𝗮𝗹 𝗳𝗿𝗶𝗲𝗻𝗱 Yash Gawde 𝗲𝘃𝗲𝗿𝘆 𝗲𝗻𝘁𝗿𝗲𝗽𝗿𝗲𝗻𝗲𝘂𝗿 𝘀𝗵𝗼𝘂𝗹𝗱 𝗹𝗶𝘀𝘁𝗲𝗻 𝘁𝗼 👇 Yes, the same Yash who’s been all over the internet for that handwritten note he passed to Nikhil Kamath at a Mumbai café. But beyond the headlines, there’s something far more powerful: 👉 𝘁𝗵𝗲 𝘃𝗶𝘀𝗶𝗼𝗻 𝗵𝗲’𝘀 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 BeHooked. Yash shares:  • Why BeHooked is more than a video tool — it’s evolving into a Media Operating System  • How he sees AI as a personal creative agent for every founder & creator  • The future of content creation where storytelling is scaled, but never loses authenticity 💡 Whether you’re a founder, a creator, or just someone curious about the future of AI — this is worth your 60 seconds. If this reel inspired you as an Entrepreneur, you will love the whole conversation on 𝗧𝗵𝗲 𝗔𝗜𝗕𝗥𝗢𝗦 𝗦𝗵𝗼𝘄, link in comments #AI #generativeAI #aibros #theaibrosshow #aibrosshow #creatoreconomy #MediaOS #founderstory #startups #theaibrosshow

39

4 Comments

In this week of WoodenScale AI's Founder of the week - We are glad to have Nithya Jagadeeswaran, who is building FinPulz. FinPulz is a modern financial platform designed to help everyday investors make smarter, safer investment decisions through curated financial products, structured learning, and trusted banking partnerships. Founders & investors, if this resonates with what you're building or investing in, feel free to connect with Nithya directly and check out FinPulz

15

1 Comment

#RAG (Retrieval-Augmented Generation) has gone from niche experiment to hot topic in a matter of months and many teams are still uncertain how to put it into practice. After our 2-part series tackling #MCP, Sunil Karkera breaks down the details of how to deal with RAG and unstructured data in part 1 of another 2-part series. In this article, he covers: - Why RAG matters and how it unlocks more reliable AI. - The challenge and what makes implementing RAG so complex. - How it works - a clear look at the moving parts. - Designing the pipeline - what choices really matter. - Why enterprises should care and what’s at stake if they don’t. Whether you’re a business leader curious about the impact, or an engineer tackling the technical details, this piece has something for you. https://lnkd.in/gaHU55D6 Soul of the Machine #AI #AgenticAI #unstructureddata

23

1 Comment

I analyzed these “great startups” from BITS Pilani and filtered them based on two categories. From my analysis, these are the Indian-headquartered companies: Oxyzo Mobile Premier League Swiggy INDmoney Axios Slice Then I looked at which of these are actually profitable: Oxyzo So from this list, Oxyzo is the only profitable Indian company that isn’t burning cash. Getting funding and posting a picture with folded hands doesn’t mean a startup is rocking like SpaceX or Tesla. Thanks to Scrapingdog Google AI Mode API for helping me analyse these startups.

12

3 Comments

Bloomberg published a «Big Take» feature on Jane Street last week. https://lnkd.in/d7VVRcCF Back in the 2000s PaulG was writing essays about ITA Software's use of Common Lisp for solving travel search, and Yaron Minsky was writing about this cool company, Jane Street, that was using OCaml for stuff in the markets. Interesting that two decades later one of these companies is in the news for all the wrong reasons.

4

OpenClaw is the Next ChatGPT Moment for India’s Top 1%. If ChatGPT was about Conversation, 2026 is about ACTION Today we are moving from "Brains in a Jar" to "Agents with Hands." OpenClaw has shattered records, becoming the fastest-growing open-source project in GitHub history because it solved the one thing that was holding AI back: Agency. For the Indian professional, this isn't just a "cool tool"; it’s the structural collapse of the "Admin Tax" on your time. Let me help you understand this better: 👇🏻 1) Unlike a chatbot that lives in a browser tab, OpenClaw is a long-running background service that lives on your machine. It doesn't just suggest a reply; it opens your WhatsApp, reads the context, drafts the response, and hits send. It’s the difference between a consultant and a chief of staff 🧠 2) Tier-1 founders like Harshil Mathur (Razorpay) are already using it to bridge the digital-physical gap. Imagine waking up to a briefing that hasn't just summarised your emails, but has already ordered your favourite breakfast because your sleep tracker showed a "low recovery" score. That is the OpenClaw + MCP (Model Context Protocol) stack in action 💪 3) In a country built on WhatsApp, OpenClaw is the ultimate equaliser. Freelancers are now running entire outreach engines - scraping LinkedIn, writing 2,700+ personalised "founder-style" emails, and updating CRMs - for as low as ₹550/mo. Small dental labs are using it to autonomously log into patient portals, download scans, and send WhatsApp updates to doctors. It’s transforming "tech-shy" businesses into AI-native operations overnight 🇮🇳 4) Originally launched as Clawdbot, then briefly Moltbot (due to trademark friction with Anthropic), it finally settled on OpenClaw. The momentum was so high that OpenAI just hired its creator, Peter Steinberger, to lead their agentic strategy, while the project itself transitions to an independent foundation 🚀 5) From a venture standpoint, we are seeing a massive shift in how we value companies. We no longer just look at headcount; we look at Output per Human. Companies using these "AI wraths" to automate 80% of their back-office aren't just leaner - they are more scalable. The value isn't in the LLM - it’s in the connectors. OpenClaw proves that "Self-Hosted" is the ultimate moat. In a world where data privacy is the new gold, the platform that lets a founder keep their "brain" on their own hardware is the one that captures the enterprise wallet 💸 The ➡️ Insight: The 2026 winner isn't the one with the best "prompts," but the one with the most efficient Agentic Density. If your team is still copy-pasting between ChatGPT and Excel, you are already losing to a competitor whose agents are doing the work while they sleep 🌙

87

7 Comments

Ops folks are our favorite kind of skeptics. They show up with exact constraints, edge cases, and “show me the log” energy, and RevenueHero has continuously evolved because of it. 2025 has been a year of massive product updates, all prioritised through conversations with some of the best minds in the ops. What we build for one deal ends up closing multiple others since our deals are primarily driven through ops buyers. Every time we launch a feature, I’ve always gone back to our calls to figure out the best way to communicate value in the words of the actual users. And finally, we’ve been able to do that. At scale. Part of our AI experiments at RevenueHero has been leveraging anonymised sales calls and combining it with product context to produce better content that speaks buyer language about the problems we solve. As part of that, here are some of the launches in 2025, and the verbatim conversations that led to these features being built: The ask (quote): “We want a better way to be able to holistically track all meetings and all meeting types being booked throughout sales.” What we built: Revamped Routing Logs A frame-by-frame replay of triggers, matching rules, weightage, fallbacks, slot pick, assignee confirmation, and downstream CRM/calendar writes. (Released: June 2025) The Ops rationale: The ask - Ops-grade observability to shorten RCA cycles and prevent “why me?” disputes with a verifiable trail. The ask (quote): “And in general, you know, if, if I can have a enrichment tool sitting on top of our, you know, bookings page or book demo page, basically. And that’s the only feature that kind of intrigued me.” What we built: Native waterfall Enrichment. (Released: May 2025) The Ops rationale: Fewer form fields, context, routing and CRM hygiene. The ask:  “the biggest, I guess, pain point in the workflow right now is, you know, scheduling meetings through HubSpot’s meeting tool when […] we need a solutions engineer or we need an analyst on the team, like multiple calendar syncing to schedule those times with clients.” What we built: Pairings (Collective Round-Robin 2.0): Pre-define fixed AE+SE duos and round-robin those pairs while presenting common availability to prospects. (Released: May 2025) The Ops rationale: Reduce reschedules and keep capacity planning predictable across roles. The ask: “We would like to get better at lead routing based on certain conditions. […] There are different use cases […] campaigns through website form fill. There is through LinkedIn form fill, LinkedIn campaigns and then generic outbound. […] Right now what’s happening is we are routing everything to one sdr.” What we built: Inbound Links — pack your full routing engine (qualification + matching + distribution) into a single link you can drop across channels. (Released: March 2025) The Ops rationale: Centralised logic to improves governance and attribution while eliminating channel-specific hacks. and more..

17

3 Comments

Startup program for e2b!! Also, why is the python tool on chat interfaces absolutely the most underrated tool. Most people don’t even know they can use python on zerotwo.ai to edit their videos and images, convert file types, do advanced math for analyzing large files of data, and more!

1

Arvind Jain said something that stuck with me: the companies that win won't be the ones that buy the most intelligence. They'll be the ones that redesign work around it. I was interviewed by Forbes last week on exactly this tension. Token consumption as a productivity signal is real but incomplete. At Starburst, we don't tokenmaxx. But I also won't put a cap on it. Teams experimenting is a good sign. The issue is when usage becomes the goal rather than the outcome. What's actually changed for us: - 60% compression in the time it takes to move an idea to production. - A third of our code is now Claude-generated. Those aren't token metrics. They're DORA metrics, deployment velocity, incident resolution time. The shift happened when we at Starburst stopped treating AI as another tool layered on top of existing workflows and started asking the more uncomfortable question: what needs to be rebuilt from the ground up? That's AI-first principles thinking. It's not about prompting better. It's about restructuring the work itself. Jensen's comment about token spend as a signal of ambition isn't wrong. But ambition without architectural change is just a bigger bill. Full piece in Forbes (link in comments). #AI #AIStrategy #EngineeringLeadership

95

4 Comments

No one had a database of historical NAVs for GIFT City Mutual Funds. So we built one. Belong - NRI savings & investments is now the first platform to offer historical NAV data and interactive NAV charts for every GIFT City Mutual Fund — something that simply didn't exist until now. Our MF Explorer pages now show complete NAV history with interactive charts. Take the Tata India Dynamic Equity Fund for example — you can track its entire NAV journey right here: https://lnkd.in/gcC_m_C2 This isn't just for one fund. We've built this database across all GIFT City Mutual Fund schemes, so you can research, compare, and track performance all in one place. This is part of our ongoing effort to bring more transparency and better tools to investors exploring GIFT City funds. More updates coming soon.

35

Decoding Emergent: How it became one of the fastest-growing AI startups? In this episode of Not Just Business, co-founder Mukund Jha talks about what drives the company, his journey, how AI may reshape jobs, and where India stands in the global AI race. https://lnkd.in/gs_9UxyY

18