ODDBIRD VC

As a founder, have you ever felt like a 𝗽𝗲𝗮𝗰𝗼𝗰𝗸 𝗶𝗻 𝗮 𝗿𝗼𝗼𝗺 𝗳𝘂𝗹𝗹 𝗼𝗳 𝗯𝗲𝗮𝗿𝘀? Or felt conventionally stereotyped? Or repeatedly asked ‘Why will you win’? Outsiders, or ‘𝙊𝘿𝘿𝘽𝙄𝙍𝘿𝙎’ as we call them, start the startup race with a handicap e.g. they are often locked outside networks of capital, or worse, judged against a ‘different’ bar by investors. We are ODDBIRD, an early-stage venture firm, born to challenge the status quo. We recognize and celebrate the outlier HUSTLE, AMBITION, RISK TAKING, and GRIT of odd birds. We want to expand the aperture of what's venture backable. So whether you are an immigrant, restarting after a failure, dropping out of college, building in a 'weird' category, or starting off solo, we want to dream alongside you, slay your demons, and help you win. We will open access to capital, customers, and insights from assisting iconic companies such as Innovaccer, Glean, Zetwerk, Apna, and many more! If you are an odd bird dreaming a new dream, hit us up at [email protected]. #oddbirdvc

Tech, Growth, and Lessons for AI-Native Startups with Dheeraj Pandey - Founder Nutanix ($20B+) and DevRev ($1B+), Mehak Aggarwal - Founder Sybill, and Riya Karumanchi - CS @ Stanford. Full video out now on our YouTube channel! https://lnkd.in/g_Qk-C3a

Few clicks from Part 1 of our "Building From The Nest" series. We had Dheeraj Pandey (Founder of 2 unicorns - DevRev ($1B+) and Nutanix($18B+)), Mehak Aggarwal (Founder, Sybill), Riya Karumanchi (MS in CS at Stanford), and 20 soon-to-be entrepreneurs share an evening filled with insights and laughs!

𝗖𝗵𝗶𝗿𝗽𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗢𝗗𝗗𝗕𝗜𝗥𝗗 𝗻𝗲𝘀𝘁 𝗹𝗮𝘀𝘁 𝘄𝗲𝗲𝗸: We’ve been chirping about the impact of AI dev tools, product design of new-age AI companies, outcomes of NIL rights, and the power of survey data. 1. AI dev tools are single-handedly changing the 𝗽𝗹𝗮𝘆𝗯𝗼𝗼𝗸𝘀 𝗼𝗳 𝘃𝗲𝗻𝘁𝘂𝗿𝗲.  𝗧𝗶𝗺𝗲 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱 𝗮 𝗽𝗿𝗼𝘁𝗼𝘁𝘆𝗽𝗲: 3-4 months → 3-4 weeks 𝗧𝗲𝗮𝗺 𝘀𝗶𝘇𝗲 𝘁𝗼 𝗵𝗶𝘁 𝘀𝗲𝗿𝗶𝗲𝘀 A: 50-member team → 5-member team  𝗖𝗮𝗽𝗶𝘁𝗮𝗹 𝘁𝗼 𝗵𝗶𝘁 $𝟭𝟬𝟬𝗠 𝗔𝗥𝗥: $100M → $10M 2. Thoughtful 𝗠𝗩𝗣 𝗱𝗲𝘀𝗶𝗴𝗻 has been a key enabler for new-gen AI companies. Most of them are 𝘀𝗲𝗹𝗳-𝘀𝗲𝗿𝘃𝗲 B2B2C products that give instant delight, enabling 𝗽𝗿𝗼𝗱𝘂𝗰𝘁-𝗹𝗲𝗱-𝗴𝗿𝗼𝘄𝘁𝗵. They integrate seamlessly into existing workflows - Ex: Cursor built on VS Code. 3. Name, Image, and Likeness (NIL) rights are unlocking massive 𝗺𝗼𝗻𝗲𝘁𝗶𝘀𝗮𝘁𝗶𝗼𝗻 opportunities for college athletes in the US. Tuition discounts, transfer fees, brand endorsements, etc. College athletes are now the 𝗺𝗶𝗰𝗿𝗼-𝗶𝗻𝗳𝗹𝘂𝗲𝗻𝗰𝗲𝗿 equivalents in sports celebrity advertising. 4. Large scale survey / 𝗽𝗲𝗼𝗽𝗹𝗲-𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲𝗱 𝗱𝗮𝘁𝗮𝘀𝗲𝘁𝘀 are significantly valuable to AI companies, investment firms, etc. They often contain the ‘why’ behind an outcome and the median of people’s opinions is more often right than not. They are a differentiator. #DevTools #Cursor #Data #AI #VentureCapital #Startups

𝗖𝗵𝗶𝗿𝗽𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗢𝗗𝗗𝗕𝗜𝗥𝗗 𝗻𝗲𝘀𝘁 𝗹𝗮𝘀𝘁 𝘄𝗲𝗲𝗸: We’ve been chirping about the rise of prediction markets, concentrations of top talent, threats to vertical AI, bottlenecks to scale in AI, and our first founder meetup! 1. Prediction markets and real money gaming are gaining traction. Consumer behaviour is increasingly changing to 𝗿𝗶𝘀𝗸-𝘀𝗲𝗲𝗸𝗶𝗻𝗴 and people are looking to build 𝗽𝗮𝘀𝘀𝗶𝘃𝗲 𝗶𝗻𝗰𝗼𝗺𝗲 𝘀𝘁𝗿𝗲𝗮𝗺𝘀. Liquidity remains a challenge but can easily become a right-to-win! 2. Any concentration of the 𝘁𝗼𝗽 𝟭% in a domain is extremely valuable - be it a club of top engineering talent, community of influencers, or a room full of founders. It is a MOAT and can be leveraged in multiple ways. 3. 𝗩𝗲𝗿𝘁𝗶𝗰𝗮𝗹 𝗔𝗜 companies will always loom under the 𝘁𝗵𝗿𝗲𝗮𝘁 of being swallowed by well-funded horizontal / broad-based AI companies if their MOAT is not strong enough (i.e. their MOAT is not 𝗱𝗮𝘁𝗮). 4. Google, Microsoft, Amazon are scrambling to increase 𝗱𝗮𝘁𝗮 𝗰𝗲𝗻𝘁𝗲𝗿 capacity citing large amounts of unmet demand. Bottlenecks in hardware infrastructure and power availability are limiting agents to 𝗔𝗜 𝘀𝗰𝗮𝗹𝗲. 5. In the first edition of our 𝗳𝗼𝘂𝗻𝗱𝗲𝗿 𝗱𝗶𝗻𝗻𝗲𝗿 𝘀𝗲𝗿𝗶𝗲𝘀, we spoke about building an 𝗔𝗜 𝗻𝗮𝘁𝗶𝘃𝗲 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 from scratch and met such passionate founders building in this space. #predictionmarkets #venturecapital #verticalai #ai

𝗖𝗵𝗶𝗿𝗽𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗢𝗗𝗗𝗕𝗜𝗥𝗗 𝗻𝗲𝘀𝘁 𝗹𝗮𝘀𝘁 𝘄𝗲𝗲𝗸: We’ve been chirping about the shift in consumer mindset towards AI, the AI MOATs, date becoming more protected, the levers to building a good AI model, and collaborative workflows. 1. The paradigm shift: for the last 4 years, AI was a “𝗴𝗼𝗼𝗱-𝘁𝗼-𝗵𝗮𝘃𝗲” in most products. Now, not having AI is almost a 𝗰𝗿𝗶𝗺𝗲. Consumers expect AI. 2. The 3 Ds -> 𝗗𝗮𝘁𝗮, 𝗗𝗶𝘀𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻, 𝗗𝗲𝘀𝗶𝗴𝗻 -> are the new MOATs for AI companies. Proprietary Data is the longest lasting MOAT, followed by Distribution and unique GTMs, followed by a well Designed (features, UI, UX) application. 3. As data becomes increasingly important, the amount of “open” data will 𝗱𝗲𝗰𝗿𝗲𝗮𝘀𝗲. Companies will become more 𝗽𝗿𝗼𝘁𝗲𝗰𝘁𝗶𝘃𝗲 about their data and companies depending solely on external data sources will 𝗱𝗶𝗲. 4. Data (quality + quantity), response time, and response quality are related. Response quality can be maintained even when training data is limited if response time is increased. S1 leveraged this. Models have 𝗺𝘂𝗹𝘁𝗶𝗽𝗹𝗲 𝗼𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗺𝗮𝘅𝗶𝗺𝗮𝘀. The use case will dictate what’s best. 5. Products that enable 𝗰𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝘃𝗲 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 𝗮𝗿𝗲 𝗱𝗶𝘀𝗿𝘂𝗽𝘁𝗶𝘃𝗲. Figma disrupted Adobe XD, Google Docs disrupted MS Word. High number of “person” prompted redesigns are an indicator of the need for a collaborative tool. Where will the next one be born? Hardware design perhaps? #data #s1 #distribution #moat #figma #venturecapital #oddbird

𝗖𝗵𝗶𝗿𝗽𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗢𝗗𝗗𝗕𝗜𝗥𝗗 𝗻𝗲𝘀𝘁 𝗹𝗮𝘀𝘁 𝘄𝗲𝗲𝗸: We’ve been chirping about AI’s shifting moats, nascency of the AI paradigm, why constraints fuel innovation, importance of superlinear growth, and our first founder dinner! 1. With AI models getting 𝗰𝗼𝗺𝗺𝗼𝗱𝗶𝘁𝗶𝘀𝗲𝗱 through open source, the MOAT will lie in the bedrock layer (𝗱𝗮𝘁𝗮, 𝗱𝗮𝘁𝗮 𝗾𝘂𝗮𝗹𝗶𝘁𝘆) and the surface layer (𝗨𝗜/𝗨𝗫, 𝗱𝗶𝘀𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻). Yes, UI/UX is not dead! 2. Paperless offices (predicted 30 years ago), autonomous cars (20 years ago) are still far from reality. Predictions like “AI agents will take over the workforce” are at least 𝟱-𝟲 𝗱𝗲𝗰𝗮𝗱𝗲𝘀 𝗮𝘄𝗮𝘆. 3. Historically, 𝗶𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻 has been preceded by 𝗰𝗼𝗻𝘀𝘁𝗿𝗮𝗶𝗻𝘁𝘀 (real or artificial). Expensive space missions -> reusable rockets, WW2 in Japan -> Toyota’s lean production system. Did US sanctions on China create a constraint that led to DeepSeek’s innovation? 4. The best businesses find extreme 𝗹𝗲𝘃𝗲𝗿𝗮𝗴𝗲 to break out from linear growth. Netflix pivoting to digital, Facebook opening its APIs, AirBnB leveraging Craigslist. Extreme risk taking ability, creativity, and open-mindedness are key 𝗳𝗼𝘂𝗻𝗱𝗲𝗿 𝗮𝘁𝘁𝗿𝗶𝗯𝘂𝘁𝗲𝘀. 5. We hosted our 𝗳𝗶𝗿𝘀𝘁 𝗳𝗼𝘂𝗻𝗱𝗲𝗿 𝗱𝗶𝗻𝗻𝗲𝗿 - an intimate gathering of founders from all stages (idea to seed to IPO) on “Building an AI-native company”. Video out soon but until then, here's a sneak peek! #ai #moat #deepseek #superlinear #risk

𝗖𝗵𝗶𝗿𝗽𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗢𝗗𝗗𝗕𝗜𝗥𝗗 𝗻𝗲𝘀𝘁 𝗹𝗮𝘀𝘁 𝘄𝗲𝗲𝗸: We have been chirping about the need for computation, advent of creative training methods, interesting sub-spaces in biotech, and how data is a strong moat. 1. 𝗗𝗲𝗲𝗽𝗦𝗲𝗲𝗸 𝗥𝟭. While much is yet to be understood, it radically challenges the belief that mega 𝗰𝗼𝗺𝗽𝘂𝘁𝗮𝘁𝗶𝗼𝗻 (read 𝗰𝗮𝗽𝗶𝘁𝗮𝗹) is necessary for AI progress. Are there other 𝗰𝗿𝗲𝗮𝘁𝗶𝘃𝗲 methods to create top performance at a fraction of the cost? 2. Even if DeepSeek R1 fast-followed / knocked off of leading models, it paints a 𝗴𝗿𝗶𝗺 picture of the 𝗱𝘂𝗿𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗼𝗳 𝘁𝗵𝗲 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝘃𝗮𝗹𝘂𝗲 of foundational model companies such as OpenAI. 3. Richard Sutton noted that 𝗯𝗿𝘂𝘁𝗲-𝗳𝗼𝗿𝗰𝗲 training + computation will outpower 𝗰𝗼𝗻𝘁𝗲𝘅𝘁𝘂𝗮𝗹 knowledge feeding in the long run - a take that we echo and one that DeepSeek proved right. This poses severe threats to the 𝗱𝗲𝗳𝗲𝗻𝘀𝗶𝗯𝗶𝗹𝗶𝘁𝘆 of vertical AI models. Are there creative training methods that could counter Sutton’s note? 4. 𝗗𝗿𝘂𝗴 𝗗𝗶𝘀𝗰𝗼𝘃𝗲𝗿𝘆, 𝗚𝗲𝗻𝗲 𝗧𝗵𝗲𝗿𝗮𝗽𝘆, and 𝗟𝗼𝗻𝗴𝗲𝘃𝗶𝘁𝘆 are spaces in biotech we’re excited about. Tailwinds - growing library of bio foundation models, compute available for research, and the commercial success of novel therapeutics (e.g. mRNA vaccines) 5. 𝗗𝗮𝘁𝗮 𝗶𝘀 𝗧𝗛𝗘 𝗠𝗢𝗔𝗧 in biotech. As models get open-sourced, teams that 𝗥𝗘𝗔𝗟𝗟𝗬 understand what data is valuable, how to manipulate data, and how to mix and match diverse data sets will succeed. #ai #data #deepseek #biotech #venturecapital

We crunched the numbers on #MahaKumbhMela and uncovered its marketing genius—delivering 𝟭𝟱𝘅 the return on $ spent compared to #Instagram! This Hindu pilgrimage will host 𝟰𝟱𝟬 𝗺𝗶𝗹𝗹𝗶𝗼𝗻 people—more than the ENTIRE population of North America! It’s like Instagram meets the Super Bowl, but on steroids.  𝗕𝘂𝘁 𝗵𝗼𝘄? • Visiting Prayagraj = Creating an Instagram profile (𝘂𝘀𝗲𝗿 𝗮𝗰𝗾𝘂𝗶𝘀𝗶𝘁𝗶𝗼𝗻) • Public dips in the river = Content consumption (𝗱𝗼𝗽𝗮𝗺𝗶𝗻𝗲 𝗵𝗶𝘁𝘀) • Mystical sages (Naga Babas) = Content creators (𝗵𝗼𝗼𝗸𝘀 𝗳𝗼𝗿 𝗿𝗲𝘁𝗲𝗻𝘁𝗶𝗼𝗻) • Akharas / sages have “ranks” = Influencers have “followers” (𝘀𝗼𝗰𝗶𝗮𝗹 𝘀𝗶𝗴𝗻𝗮𝗹𝗶𝗻𝗴) • Devotees bring friends & family = Users bring (𝗻𝗲𝘁𝘄𝗼𝗿𝗸 𝗲𝗳𝗳𝗲𝗰𝘁𝘀) • PM Modi’s presence = MJ at the Super Bowl halftime (𝗺𝗮𝗿𝗾𝘂𝗲𝗲 𝗮𝗱𝗼𝗽𝘁𝗶𝗼𝗻) • Stalls and banners = promoted stories and posts (𝗮𝗱𝘃𝗲𝗿𝘁𝗶𝘀𝗶𝗻𝗴 𝗲𝗰𝗼𝗻𝗼𝗺𝘆) • Temporary tents and infrastructure = App UI/UX (𝘀𝗲𝗮𝗺𝗹𝗲𝘀𝘀 𝘂𝘀𝗲𝗿 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲) 𝗧𝗵𝗲 𝘃𝗶𝗿𝗮𝗹𝗶𝘁𝘆 𝗹𝗼𝗼𝗽: Eyeballs →enjoyable experience → bring friends → consume in public → FOMO -> new consumers → more eyeballs! 𝗦𝗼 𝘄𝗵𝗮𝘁? Maha Kumbh → Revenue: $23B, Cost: $1B → Return on $ spent = 𝟮𝟯𝘅 Instagram→ Revenue: $135B, Cost: $87B → Return on $ spent = 𝟭.𝟱𝘅 A stunning contrast. 𝗬𝗲𝘀, 𝘁𝗵𝗲 𝗠𝗮𝗵𝗮𝗞𝘂𝗺𝗯𝗵𝗠𝗲𝗹𝗮 𝗼𝗰𝗰𝘂𝗿𝘀 𝗼𝗻𝗰𝗲 𝗶𝗻 𝟭𝟰𝟰 𝘆𝗲𝗮𝗿𝘀 𝘀𝗼 𝗽𝗲𝗻𝘁 𝘂𝗽 𝗱𝗲𝗺𝗮𝗻𝗱 𝗰𝗼𝘂𝗹𝗱 𝗯𝗲 𝗮 𝗳𝗮𝗰𝘁𝗼𝗿? A close comparable is the Olympics, held every 4 years. The recent Paris Olympics generated a revenue of $6B with a return on $ spent of just 𝟬.𝟱𝘅. The comparison isn’t perfect, but there are many powerful takeaways! My personal favourites: 𝟭. 𝗢𝗳𝗳𝗹𝗶𝗻𝗲 𝘀𝗵𝗼𝘂𝗹𝗱𝗻’𝘁 𝗯𝗲 𝘀𝗰𝗮𝗿𝘆 - Online channels can quickly plateau and lose marketing efficiency once saturated. 𝟮. 𝗕𝗿𝗮𝗻𝗱𝘀 𝗮𝗿𝗲 𝗲𝗻𝗱𝘂𝗿𝗶𝗻𝗴 - The Kumbh Mela is CENTURIES old. Long-term brand loyalty drives sustained growth and impact. 𝟯. 𝗩𝗶𝗿𝗮𝗹 𝗹𝗼𝗼𝗽𝘀 - Present everywhere and if harnessed, yields incredible marketing efficiency ODDBIRD VC #venturecapital #india #startups

𝗖𝗵𝗶𝗿𝗽𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗢𝗗𝗗𝗕𝗜𝗥𝗗 𝗻𝗲𝘀𝘁 𝘁𝗵𝗶𝘀 𝘄𝗲𝗲𝗸: We have been chirping about AI agent lifecycle, right personas to target in healthcare, complex coordination, the death of UI, and food preferences for founders. 1. ChatGPT was unbundled in 23/24 and vertical agents were born. These agents are 𝗲𝘃𝗼𝗹𝘃𝗶𝗻𝗴 𝗹𝗶𝗸𝗲 𝗵𝘂𝗺𝗮𝗻𝘀 - birth -> exposure to information -> independent thought. Next steps are 𝗰𝗼𝗺𝗺𝘂𝗻𝗶𝗰𝗮𝘁𝗶𝗼𝗻 (with other agents and the outside world) and self-driven 𝗲𝗱𝘂𝗰𝗮𝘁𝗶𝗼𝗻 (DIY training) 2. As tasks shift to AI agents, reliance on traditional UI/UX may decline. Moreover, we are able to do more than ever with just voice, images, and videos. With 𝗳𝗲𝘄𝗲𝗿 𝗱𝗶𝗴𝗶𝘁𝗮𝗹 𝗮𝗰𝘁𝗶𝗼𝗻𝘀 required, complex interfaces could become obsolete. Is this the 𝗱𝗲𝗮𝘁𝗵 𝗼𝗳 𝗨𝗜? 3. Primary care providers are great GTM channels for health-techs but offer 𝗹𝗼𝘄 𝗺𝗼𝗻𝗲𝘁𝗶𝘇𝗮𝗯𝗶𝗹𝗶𝘁𝘆. Specialty departments, like 𝗢𝗻𝗰𝗼𝗹𝗼𝗴𝘆, have higher revenues per patient and longer tails - personalized treatments for each patient. This means higher margins, and hence fatter wallets to take bites out of. 4. In a world of information overload, 𝗰𝗼𝗺𝗽𝗹𝗲𝘅 𝗰𝗼𝗼𝗿𝗱𝗶𝗻𝗮𝘁𝗶𝗼𝗻 is a MOAT - “How does one align different elements to create value for end users?”. This is all the more important in AI as the ability to do more with less is prevalent but 𝗲𝘅𝘁𝗿𝗮𝗰𝘁𝗶𝗻𝗴 𝗿𝗲𝗮𝗹 𝘃𝗮𝗹𝘂𝗲 𝗶𝘀 𝗮 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲. 5. We’re debating dinner choices for our 𝘂𝗽𝗰𝗼𝗺𝗶𝗻𝗴 𝗱𝗶𝗻𝗻𝗲𝗿 𝘀𝗲𝗿𝗶𝗲𝘀. Burgers? Salads? A freshly baked thin-crust pizza, oozing with 𝗺𝗲𝗹𝘁𝗲𝗱 𝗺𝗼𝘇𝘇𝗮𝗿𝗲𝗹𝗹𝗮 and fragrant basil? Well, we’re clearly biased and we could use some help. Founders, let us know what fits your cravings and we’ll serve it up! #ai #agents #oncology #healthcare #venturecapital #pizza