System3

My unpopular opinion for a Saturday: The hardest skill for an AI founder right now isn't speed. It’s 𝘀𝗲𝗹𝗲𝗰𝘁𝗶𝘃𝗲 𝗶𝗴𝗻𝗼𝗿𝗮𝗻𝗰𝗲. I was speaking to a fellow founder this week who wanted to pivot their entire roadmap because of a rumor about GPT-5. Last month, they wanted to pivot because of Gemini 2.0. I noticed the 𝙋𝙞𝙫𝙤𝙩 𝙏𝙧𝙖𝙥. In an industry moving this fast, it is terrifyingly easy to let the news cycle dictate your strategy. You feel like if you aren't reacting to every paper and product drop, you're dying. But here is the contrarian truth I’m reminding myself of at my work every day: 𝗬𝗼𝘂𝗿 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿𝘀 𝗱𝗼 𝗻𝗼𝘁 𝗰𝗮𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗺𝗼𝗱𝗲𝗹 𝗯𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸𝘀. They don't care if the new context window is 1 𝘮𝘪𝘭𝘭𝘪𝘰𝘯 or 10 𝘮𝘪𝘭𝘭𝘪𝘰𝘯 tokens. They care if you solved the headache they had on Tuesday. If your product value collapses every time Sam Altman tweets, you never had a product; you had a feature. The winners of 2026 won't be the ones who reacted the fastest to the news. They will be the ones who ignored the noise long enough to build something that actually sticks. Are you building for the technology, or are you building for the customer? It’s getting harder to tell the difference.

The narrative is "𝘌𝘹𝘱𝘢𝘯𝘥𝘪𝘯𝘨 𝘈𝘤𝘤𝘦𝘴𝘴." The reality is the gentrification of truth. OpenAI just announced the inevitable: Ads are coming to ChatGPT, alongside the rollout of the $8/month 'ChatGPT Go' plan (which they interestingly validated in India first). The pitch is standard tech economics: 𝘈𝘥𝘴 𝘴𝘶𝘣𝘴𝘪𝘥𝘪𝘻𝘦 𝘵𝘩𝘦 𝘤𝘰𝘮𝘱𝘶𝘵𝘦 𝘴𝘰 𝘵𝘩𝘦 𝘸𝘰𝘳𝘭𝘥 𝘨𝘦𝘵𝘴 𝘢𝘤𝘤𝘦𝘴𝘴 𝘵𝘰 𝘪𝘯𝘵𝘦𝘭𝘭𝘪𝘨𝘦𝘯𝘤𝘦. Sure. But as builders, we need to talk about the hidden cost of this architecture. We are witnessing the birth of "Two-Tiered Intelligence." • Tier 1 (The Enterprise/Pro): 𝗣𝗮𝘆𝘀 𝗳𝗼𝗿 𝗽𝗿𝗶𝘃𝗮𝗰𝘆. They get a research assistant optimized for accuracy. • Tier 2 (The Public): 𝗣𝗮𝘆𝘀 𝘄𝗶𝘁𝗵 𝗮𝘁𝘁𝗲𝗻𝘁𝗶𝗼𝗻. They get a research assistant that has a side-deal with a sponsor. OpenAI promises "𝘈𝘯𝘴𝘸𝘦𝘳 𝘐𝘯𝘥𝘦𝘱𝘦𝘯𝘥𝘦𝘯𝘤𝘦." I want to believe them. But business models are destiny. When your revenue comes from an advertiser, the optimization function of the model inevitably shifts. It moves from "Save the user time" to "Maximize engagement." It moves from "Hard Truths" to "Brand-Safe Responses." If true AI Literacy means the ability to discern truth from hallucination, we now have a much harder challenge: discerning truth from persuasion. My end thought remains the same: Is "Sponsored Intelligence" better than "No Intelligence"? Or are we creating a digital class divide where only the wealthy can afford the unbiased truth? #OpenAI #AIEthics #TechStrategy #FutureOfWork

The era of "General Purpose AI" might be closing. OpenAI just signaled the start of the Verticalization Race. OpenAI just launched OpenAI for Healthcare. Most people see this as a standard product launch. I see it as the most significant strategic pivot of the last 24 months. For three years, the industry thesis was: 𝘛𝘳𝘢𝘪𝘯 𝘰𝘯𝘦 𝘮𝘢𝘴𝘴𝘪𝘷𝘦 𝘮𝘰𝘥𝘦𝘭 𝘰𝘯 𝘵𝘩𝘦 𝘦𝘯𝘵𝘪𝘳𝘦 𝘪𝘯𝘵𝘦𝘳𝘯𝘦𝘵 𝘢𝘯𝘥 𝘭𝘦𝘵 𝘪𝘵 𝘥𝘰 𝘦𝘷𝘦𝘳𝘺𝘵𝘩𝘪𝘯𝘨. That worked for writing emails and coding Python. It failed for diagnosing rare diseases or underwriting complex loans. Why? Because the 𝗽𝘂𝗯𝗹𝗶𝗰 𝗶𝗻𝘁𝗲𝗿𝗻𝗲𝘁 𝗶𝘀 𝘁𝗮𝗽𝗽𝗲𝗱 𝗼𝘂𝘁. You cannot train a world-class doctor using only Reddit threads and Wikipedia articles. To reach the next level of utility, AI companies are desperate for "Expert Trajectories." They need to see how a doctor thinks, not just what a textbook says. That data doesn't live on the web. It lives inside walled gardens (EHRs). This is the new playbook for 2026: OpenAI is pivoting from a Horizontal Platform (selling GPT-4 to everyone) to a Vertical Partner (building deep silos in Med, Law, and Finance). They are trading specific software utility for access to specific expert reasoning. What this means for Founders & Operators: 𝗧𝗵𝗲 "𝗪𝗿𝗮𝗽𝗽𝗲𝗿" 𝗶𝘀 𝗶𝗻 𝗧𝗿𝗼𝘂𝗯𝗹𝗲: If you built a "ChatGPT for Doctors" startup, the platform just entered your territory. 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗶𝘀 𝗞𝗶𝗻𝗴: The value has shifted from the "Model" (which is becoming a commodity) to the "Proprietary Data" (which is the only way to build a moat). The future isn't a single god-like AI that knows everything. It is a team of deep specialists, trained on data you can't Google. We are moving into a world where your Data Strategy is the only thing protecting your business from the platforms. If you are building in a deep vertical (EdTech, Legal, Finance) and thinking about how to defend your moat, let’s exchange notes. #AIStrategy #HealthTech #VerticalAI

The marginal cost of knowledge has hit zero. The value of practice is skyrocketing. For the last decade, the EdTech playbook was simple: Digitize the textbook. Put the lecture on video. Gate the access. GenAI broke that model. When expertise is free, instant, and synthesized, selling "access to information" is a race to the bottom. 💯 Building System3 taught me that the new scarcity isn't the answer, it’s the context. We stopped trying to compete with the internet on "what is the right framework?" and started competing on "how do I apply this framework in a messy, high-pressure human situation?" We moved from consumption to simulation. The future of workforce development isn't a better library. It’s a scalable flight simulator for soft skills. If your L&D strategy is still based on video completion rates rather than simulation outcomes, you aren't building skills. You're just hosting files. If you’re working on AI, edtech, or skills-building and want to brainstorm ideas or pressure-test a strategy, I’m always happy to be in the room. #EdTech #FutureOfWork #AI

India will have the world’s largest youth population. A demographic advantage only matters 'when' it becomes a skills advantage. I’ve spoken to numerous students across schools, colleges and early careers. They are curious, ambitious and more than capable. But they’re not getting the structured, future learning they deserve. The potential is there? The access? Not so much. India DOES not need to catch up in AI tools. We need to start from scratch. AI literacy is the need of the hour. If we get the foundation right, we can build the strongest talent pipeline in the world. Here’s how we can go about building it 👇 1/ Start early 2/ Make AI a literacy. 3/ Teach thinking, not memorizing.  4/ Build creators, not just consumers. India has the talent ✅ India has the ambition ✅ What we need right now is the right infrastructure and support for AI learning. ✅ Think about this, If we make AI education accessible, rigorous and hands on today, We don’t just participate in the AI global wave. We lead it. -Anshika What’s the biggest barrier to AI literacy in your view? Let me know in the comments. #AI #education

AI tools are evolving fast. Classrooms and institutions are not. The gap seems not to be in tech, but in the current mindset and methods. When I started introducing AI concepts to students and educators, I assumed the hardest part would be explaining the tech. I was wrong. Turns out, the real challenge was to help people un-learn the outdated ways of learning. But Education still treats AI like a subject. -> A side topic -> A nice to have -> An extra basic skill AI is not the chapter though. It is the new learning environment itself. What actually works looks like this: 1/ Early exposure 2/ Mentorship driven learning 3/ Project first thinking  4/ Continuous iteration AI can accelerate learning. But without structure, guidance and depth, students only learn about AI. Not how to think with it Not how to build with it The future will reward those who learn, build and think with AI. -Anshika What mindset shifts do you think educators or leaders still need to make? Let me know in the comments. #AI #education

China’s student interest in AI degrees rose 88% year on year. While interest in the US fell 25%. Talent pipelines, and not just AI tools will decide who leads and wins the AI race. When I look across data, I see two very different but constructive stories. One from the US, and the other from China, and here are 3 lessons every organization or educator can act on: 👇 1/ Build the pipeline early: -> China seems to be doing this: Aggressive up-skilling, national plans to “self rely” on AI talent. -> US leads in research and investment, but talent funnel seems to be lagging behind 2/ Link education, industry and policy in one flow -> China leverages their government policy, universities and industry to grow AI talent rapidly. -> The USA has innovation but less co-ordination across the pipeline. 3/ Treat literacy as a capability  -> Both nations highlight that having skills alone is not enough. -> For professionals, learners, and institutes, the goal is creating daily competence in AI thinking, working, and creating. Here’s the harsh truth ❗ The next wave of AI value will not come from countries with the most models or data sets. It will come from the place who has the most trained and skilled people. If you’re an educator, ask yourself this: Are you building the next AI tool? Or are you building the next wave of talent? 🤔 -Anshika What do you think matters most when building an AI talent pipeline? Let me know in the comments. #AI #education

According to new data, 86% of education institutions worldwide are already using generative AI, the highest adoption rate across any industry. 👏 💯 Students are learning with it. 💯 Educators are teaching with it. But not everyone understands how to use it effectively. Swipe through to see what 2025 data reveals about how AI is truly reshaping classrooms. 💡 #AI #Education #AIInEducation

Harvard isn’t asking "Should AI be used in education?" They're rather telling us how to use it responsibly, rigorously, and creatively. Because AI literacy isn't optional anymore. It's the new foundation of academic excellence. Swipe to see what the world’s top university is doing and why Indian schools must move now.

AI has definitely made building MVP’s easier. But let's get one thing straight. The new narrative that programming is dead and that "English is the new coding language”? I heard this on a well known podcast on youtube and it made me wonder why a lot of fear mongering related to AI is happening at the moment. This is not just wrong, it’s baseless! It creates nothing but fear. And fear is the last place you want to start learning AI from! If you’re new to AI, the goal isn't to change your skill. It should be to enhance your existing skills. And certainly AI can help you become faster, sharper, and more creative. Here’s how I would approach AI if you’re just starting: 1/ Start with what you hate doing. -> Hate writing email follow-ups? Hate sorting data? Start there. -> Let AI take over the tasks you dread. 2/ Build one good prompt. -> Spend 10 minutes crafting one perfect prompt for a task you do every day. -> Use it over and over. -> One great prompt is better than 100 bad ones. 3/ Ask "why," not just "what." -> When AI gives you an answer, ask "Why did you suggest this?" or "What's a better alternative?" -> This is how you learn from it, not just copy from it. So the next time you hear someone say that AI is here to replace you, remember: The ones who use AI will replace the ones who don't Stay smart. Stay curious. Stay relevant. -Anshika #AI #Technology #Education