gAI ventures

𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐡𝐚𝐝 𝐫𝐞𝐦𝐚𝐢𝐧𝐞𝐝 𝐬𝐭𝐚𝐠𝐧𝐚𝐧𝐭 𝐟𝐨𝐫 𝐟𝐚𝐫 𝐭𝐨𝐨 𝐥𝐨𝐧𝐠. apart from schools and colleges, learners started using🍎 internet and 📺 youtube. Despite the internet being the best classroom and YouTube the ultimate on-demand tutor, and some edutech tutor platforms there’s still something missing in education—true personalization, measuring progress and meaningful engagement. 𝐃𝐚𝐲 9 𝐨𝐟 𝐀𝐈, where I promised I’ll be sharing one compelling story per day—each packed with real-world data and insights I hv learned throughout 2024. So how do we deliver more customization, method (intense step-by-step courses), measure and relevance. Generative AI might hold the key to unlocking a future of smarter, more aligned learning experiences 𝑲𝒉𝒂𝒏 𝑨𝒄𝒂𝒅𝒆𝒎𝒚: has implemented a pilot program to create “Khanmigo,” an AI-powered tutoring chatbot designed to assist students with personalized learning support. it uses GPT-4 and some 100,000 students and teachers piloted Khanmigo this past academic year in schools nationwide, helping to flag any hallucinations the bot has and providing tons of student-bot conversations for DiCerbo and her team to analyze. During the 2023-24 school year, more than 221.2K individuals used Khanmigo. 𝑹𝒊𝒊𝒊𝒅: This South Korean company has developed an AI-powered tutoring system: Their AI tutor teaches, assesses, recommends learning paths, and analyzes performance. It creates personalized study plans to improve outcomes on exams and standardized tests. Riiid was recently included in the CB Insights AI 100 list of companies transforming the edtech industry. 𝑫𝒖𝒐𝒍𝒊𝒏𝒈𝒐: The language learning platform now has AI-driven features like mistake explanations and practice conversations to enhance user experience. users can ask to “Explain My Answer” and “Roleplay,” to provide detailed feedback and interactive scenarios for learners. estimated that a single digit % of the 50 mn (MAUs) learners have tried this feature tht utilizes GPT-4 to power its “Duolingo Max” subscription tier. Their stock has tripled since the IPO 𝑫𝒓𝒆𝒂𝒎𝑩𝒐𝒙 𝑳𝒆𝒂𝒓𝒏𝒊𝒏𝒈: This platform uses AI algorithms to provide adaptive math lessons for K-8 students: It analyzes how students approach problems to determine learning styles. The AI tracks progress in real-time and tailors the curriculum accordingly. DreamBox's AI can predict future proficiency based on ongoing assessments These platforms demonstrate how generative AI is being integrated to enhance personalization, provide intelligent tutoring, and improve overall learning experiences in the edtech sector. a lot of other edutech platforms are still tinkering around. offering ChatGPT plugin to help users discover learning materials. they need to try harder :)

Demis Hassabis, CEO of DeepMind, reflects on AI's evolution, inspired initially by brain architectures like neural networks and reinforcement learning. As we move from scaling massive foundation models to applying AI across sciences, Hassabis envisions a future where AI helps us analyze and understand the human brain itself, completing a fascinating loop. Speaking at the Royal Palace in Stockholm with Nobel laureates from physics, chemistry, medicine, and economics, he highlighted how AI could redefine neuroscience and unlock the mysteries of human cognition. #AI #DeepMind #Neuroscience #nobelprize #google #deepmind

Why did the AI lawyer get disbarred? Because it couldn’t pass the Turing Testimony! 😅 ok jokes apart. AI won’t just speed up the legal system — it will revolutionise it 𝑫𝒂𝒚 7 𝒐𝒇 𝑨𝑰 (skipped one day. I was travelling) I will cover 3 companies today as exhibits: 1) Thomson Reuters acquired 𝐂𝐚𝐬𝐞𝐭𝐞𝐱𝐭, 𝐚 𝐥𝐞𝐠𝐚𝐥 𝐭𝐞𝐜𝐡 𝐀𝐈 𝐜𝐨𝐦𝐩𝐚𝐧𝐲, 𝐢𝐧 𝐀𝐮𝐠𝐮𝐬𝐭 2023 𝐟𝐨𝐫 $650 𝐦𝐢𝐥𝐥𝐢𝐨𝐧. Must watch Casetext founder YC podcast interview. but basically Casetext is renowned for its advanced AI/ML solutions tailored for legal professionals, notably its AI legal assistant, CoCounsel, powered by GPT-4. CoCounsel streamlines tasks such as document review, legal research memos, deposition preparation, and contract analysis, significantly enhancing efficiency and the quality of legal representation. 2) LexisNexis ♨️ LexisNexis has introduced Lexis+ AI, a generative AI solution designed to assist with tasks like drafting legal documents and summarizing case law, featuring linked, verifiable legal citations to ensure accuracy.  💡  The platform enhances legal research by providing conversational search capabilities, allowing practitioners to obtain precise information efficiently Sean Fitzpatrick, CEO of LexisNexis North America, UK, and Ireland. “We are working closely with customers to integrate our leading generative AI tools into their workflows to drive unprecedented efficiencies and quality for both law firms and corporate legal departments.” 3) 𝐇𝐚𝐫𝐯𝐞𝐲 𝐀𝐈 is a generative AI platform tailored specifically for the legal profession. It assists legal professionals with tasks such as contract analysis, due diligence, litigation support, and regulatory compliance. By leveraging advanced AI capabilities, Harvey AI helps generate insights, recommendations, and predictions based on data, enabling lawyers to deliver faster and more cost-effective solutions for their clients. They have raised a total of $206 million over four funding rounds Challenges: well, there are many. but one that I would like to point out based on my research - 80% of Fortune 1000 execs expect law firms to reduce bills due to AI efficiency gains. But only 9% have actually told their law firms this. Classic case of everyone thinking it but nobody saying it. Kushal Prakash Vineet Mohan Vijay R. Harvey Casetext, Part of Thomson Reuters Caveat: AI generated image. these are not logos

𝐖𝐡𝐚𝐭 𝐢𝐬 𝐜𝐨𝐦𝐦𝐨𝐧 𝐛𝐞𝐭𝐰𝐞𝐞𝐧 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐀𝐈, Gen Z, chatbot 𝐚𝐧𝐝 𝐂𝐨𝐦𝐞𝐝𝐢𝐚𝐧𝐬? 𝑫𝒂𝒚 4 𝒐𝒇 𝑨𝑰 (yes I haven't forgotten abt the 18 days challenge yet ;)): Cleo, 8-year old personal finance chatbot, has demonstrated remarkable growth and success through its strategic use of 𝐠𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐀𝐈 𝐚𝐧𝐝 𝐋𝐋𝐌𝐬, particularly over the past few months. While other personal finance chatbots from the 2015-2016 messaging boom pivoted, 𝐂𝐥𝐞𝐨 𝐝𝐨𝐮𝐛𝐥𝐞𝐝 𝐝𝐨𝐰𝐧 𝐨𝐧 𝐀𝐈 𝐜𝐡𝐚𝐭. Around a third of Cleo’s 300+ team works on its chatbot functionality. Now, Sacra estimates that: ♨️ Cleo is at $150M annualized recurring revenue (in Nov'24), up 97% YoY 🚀 with LLMs more than 2x’ing their engagement and 3x’ing their revenue from upselling high-margin financial products. So what do they do actually? 🍎 By accessing your income, bills and spending, and combining them with your life goals, Cleo’s AI is built to encourage better financial habits.  👍 Credit Analysis and Budget Planning: Using open banking data, Cleo integrates AI to create personalized financial management plans. 👉 The adoption of LLMs has more than doubled user engagement, leveraging the chatbot’s “personality-driven” design, like its humorously bold 😆 “Roast Mode” that critiques spending habits. you have to give it to them for their very innovative and creative use of GenAI and their wittiness. Check out this cool video by Team Cleo on "how they use AI" 👏 Barney Hussey-Yeo https://lnkd.in/guYDF3yA 🏦 Compliance & Risk mgmt: Cleo spends heavily on annotating chatbot conversations and fine-tuning LLM models, ensuring compliance and managing potential risks from AI conversations. and for particularly sensitive content, it won’t use LLMs and instead will use traditional machine learning to audit conversations. oh but what is common between Generative AI, GenZ and Comedians? its Cleo. 1) Cleo employs ten *comedians* to ensure a culturally relevant and witty user experience, combining LLMs with traditional ML to deliver unique and personalized interactions 2) Cleo’s ability to blend personality-driven generative AI with tailored financial solutions highlights its success in engaging *Gen Z* and Millennial users while boosting its revenue streams and scaling globally.

🚀 Day 2 𝐰𝐡𝐞𝐫𝐞 𝐈 𝐬𝐡𝐚𝐫𝐞 𝐚𝐛𝐨𝐮𝐭 𝐀𝐈-𝐟𝐢𝐫𝐬𝐭 𝐞𝐫𝐚 𝐜𝐨𝐦𝐩𝐚𝐧𝐢𝐞𝐬. As promised over the next 18 days leading up to the New Year, I’ll be sharing one compelling #GenAI story per day—each packed with real-world data and insights. 70% to 90% data in enterprises is unstructured data. most of the software/SaaS which was built over the last 20 years was all about structure data - required structured input into software, learning curve, structured dashboards, specific queries it could carry and such. we are finally able to deal with unstructured data directly and get to the answers. reduce time, cost and increases insights. Glean and the Rise of AI-First Enterprise Search ++ 2019: Glean started as a humble NLP search engine for documents and apps, helping businesses answer questions like “What are our Q4 targets?” fast. But when large language models (LLMs) entered the picture, the company skyrocketed to an AI-powered future. Today, Glean stands tall with: 🚀 $100𝐌 𝐀𝐑𝐑 (𝐍𝐨𝐯𝐞𝐦𝐛𝐞𝐫 2024), 𝐠𝐫𝐨𝐰𝐢𝐧𝐠 203% 𝐘𝐨𝐘 ♨️ Valuation: $4.6B (46x revenue multiple) Glean Assistant: An AI copilot tapping proprietary internal data to do more than search—summarize, chat, and generate content 🔍 The Landscape: This isn’t just a story about Glean. It’s part of a brewing AI battle royale featuring: AI-first upstarts: Writer ($47M ARR, 200% YoY), Hebbia ($13M ARR, 160% YoY), Sana AI ($20M ARR, 300% YoY) acc to Sacra. Legacy productivity titans: Zapier ($310M ARR), Airtable ($375M ARR), Notion ($250M revenue in 2023). Big-tech giants: Microsoft Copilot, Google Workspace with Gemini, Dropbox Dash. 💡 The Opportunity: Whether via vertical integration (LLM + app layer, like Writer with Palmyra) or a compound approach (using multiple LLMs, like Glean or Dropbox Dash), the race to own enterprise AI is accelerating. For Glean and its AI-driven peers, the future of workflows, productivity, and enterprise innovation has only just begun. Talk to us gAI ventures if you want an AI enterprise search capability connected to all your systems/data or data lakehouse. we have built a very effective system digital memory + Enterprise search replete with doc processing engine, etc. (DM me or Kushal Prakash)

The Future of Science: Democratized by AI and Quantum Innovation in AI and quantum technology is reshaping science, making it more accessible to the masses. While elite science will remain at the forefront, the computational power of these technologies will empower millions - if not billions, to contribute to breakthroughs in fields like material science and biology. Even non-scientists will find themselves pushing the boundaries of discovery and productivity. #science #aiinnovation #quantumphysics #scientist #engineers

Google's Gemini 2.0: Your Ultimate Assistant & Gaming Buddy Meet Gemini 2.0, the AI that revolutionizes your digital experience. Whether strategizing in games, remembering your to-dos, or searching the web in real-time to offer tailored advice, Gemini is the ultimate multitasking companion. Redefine productivity and play with AI that truly understands you. #AI #Gemini2 #Google #AIAssistant #ArtificialIntelligence

Could AI teach terrorists to build bioweapons, hackers to launch mass cyberattacks, or even reprogram & reproduce itself? Anthropic’s Red Team is racing to find out. Led by Logan Graham and supported by experts like Newton Cheng, Daniel Freeman, and Anjali Gopal, is working tirelessly to ensure these powerful systems remain safe and beneficial. In a recent series of evals, Anthropic tested its Claude Sonnet 3.5 model for risks including: 𝗖𝘆𝗯𝗲𝗿𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝘁𝗵𝗿𝗲𝗮𝘁𝘀: Testing capabilities for hacking, exploiting vulnerabilities, and bypassing cryptographic safeguards. 𝗕𝗶𝗼𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝘁𝗵𝗿𝗲𝗮𝘁𝘀: Evaluating the AI’s ability to generate information on chemical or biological weapon development. 𝗔𝘂𝘁𝗼𝗻𝗼𝗺𝘆 𝗿𝗶𝘀𝗸𝘀: Examining the AI’s capacity to complete complex tasks independently, including reprogramming itself or other AIs. After weeks of rigorous testing, Sonnet 3.5 was deemed safe for release, maintaining an 𝗔𝗜 𝗦𝗮𝗳𝗲𝘁𝘆 𝗟𝗲𝘃𝗲𝗹 2 (𝗔𝗦𝗟-2) classification. Here’s what the tests revealed: 1. It could exploit simple vulnerabilities but struggled with advanced hacking or cryptography. 2. Attempts to generate dangerous biological or chemical instructions failed due to inaccuracies. 3. It performed programming tasks faster than novices but remained far from expert-level capabilities. As Graham aptly puts it, “The first thing that’s at stake is catastrophe.” By identifying vulnerabilities before release, Anthropic is shaping a safer AI future. #AI #Cybersecurity #Ethics #Technology

𝐀𝐈 𝐢𝐬𝐧’𝐭 𝐣𝐮𝐬𝐭 𝐭𝐞𝐜𝐡; 𝐢𝐭’𝐬 𝐚 𝐫𝐞𝐯𝐞𝐧𝐮𝐞 𝐚𝐧𝐝 𝐞𝐧𝐠𝐚𝐠𝐞𝐦𝐞𝐧𝐭 𝐦𝐮𝐥𝐭𝐢𝐩𝐥𝐢𝐞𝐫. Over the next 18 days leading up to the New Year, I’ll be sharing one compelling story per day—each packed with real-world data and insights. 𝑫𝒂𝒚 1 𝒐𝒇 𝑨𝑰: With its acquisition of foundation model company Leonardo.Ai --> Canva has seen more than 11 billion uses of its AI tools and is reinventing itself as an AI company. but how? Canva shows what AI can do for engagement + revenue: 👏 800M monthly uses of its AI tools (+700% YoY). 🍎 Teams plan price went from $300 to $500/year, increasing ACV by 66% largely due to AI engagement leverage 🚀 MAUs up 49% YoY to 220M as per Sacra 🫰 Operating profitability for the past 7 years with a highly efficient product-led growth (PLG) sales motion gives Canva the free cash flow to invest heavily into AI and enterprise sales simultaneously without large, repeated capital fundraises 🚀 ARR has surged to $2.55B (+44% YoY) in Sep 24 This isn’t about adding AI features—it’s about redefining the business with AI at the core. From acquiring Leonardo AI to launching 120+ AI plugins, Canva has used AI to boost engagement, expand ARPU, and even grow its valuation (now at $32B). AI isn’t a side feature for Canva—it’s becoming the core as they go for IPO next yr or 2026. In this 18-days series I will share Growth lessons from the AI-first era for companies who are taking advantage of this technology shift. Talk to us gAI ventures if you want to experiment with AI (DM me) Kushal Prakash Pradhumn Vijayvargiya Vineet Mohan Melanie Perkins OpenAI Syed Musheer Ahmed som sagar Simon Law Vijay R.

Seamless Productivity with FastTrackr AI Seasoned investors and founders Amit Goel and Kushal Prakash discuss a universal pain point for founders: managing busy schedules, emails, and meetings manually. Inspired by their own challenges. Introducing FastTrackr, the ultimate 24/7 executive assistant that simplifies email drafting, meeting scheduling, and note-taking - all from the comfort of WhatsApp. Designed to save you time and enhance focus, it’s like having a personal assistant in your pocket. Join thousands of professionals already unlocking their full potential with FastTrackr. #ai #productivity #WhatsApp #assistant #imessage #FastTrackrAI #worksmarter