The Ravit Show

AWS re:invent is always so much fun. I got a chance to try my hands on Trivia that happened at the AWS Village at the Data & Analytics booth. If you are at the conference go ahead and try the trivia it’s so much fun and you obviously win gifts 🥳 Good work by Analytics team 👌🏻

Day 1 -- Here at AWS re:Invent! We had a blast hosting, Pratik Das from Data & Analytics Team at AWS Village. Ravit and Pratik spoke about catalog federation for Apache Iceberg tables in the AWS Glue Data Catalog, Customer stories and more! You can learn more about this announcement here - https://lnkd.in/dUZ9sr49 If you are at AWS re:Invent don’t forget to meet Data & Analytics Team at AWS Village #data #ai #awsreinvent #aws #agents #awspartners Amazon Web Services (AWS) AWS Partners Abbas Makhdum Debika D. #awscompetencypartners #aws #theravitshow

Day 1 at AWS re: Invent — At Data & Analytics booth today, I had a blast chatting with Tej Luthra and Arturo Bayo about AI with Amazon SageMaker, what are customers excited about, use cases, and more!!!! If you are at AWS re:Invent don’t forget to meet Data & Analytics Team at AWS Village #data #ai #awsreinvent #aws #agents #awspartners Amazon Web Services (AWS) AWS Partners Abbas Makhdum Debika D. #awscompetencypartners #aws #theravitshow

Join us for a live look at what to expect at this years re:invent conference live from Las Vegas

Don’t forget to join, Ravit Jain and Zack Wenthe 🧠📢 from Vegas!!!!

(For MongoDB for Startups) Most founders do not have a model problem. They have a data and execution problem. That is why I like what MongoDB is doing with its startup program. Learn more and apply here: https://fandf.co/3XEJOrc MongoDB for Startups gives early stage teams a real boost around three things that always hurt in the first few years: - Infrastructure cost - Technical depth around data and AI - Visibility and support from a trusted platform The program is built for companies that want to: - Use MongoDB Atlas as their main data platform - Ship AI and data driven features fast - Stay flexible while they search for product market fit You get access to credits, best practices, and support so you can focus on building instead of wrestling with basic plumbing. For AI builders, this means you can spend more time on the product and less time reinventing storage, queries and data access. I see more and more strong startups standardizing on MongoDB early, then scaling without having to rip and replace later. That is a big advantage. If you are a founder in my network who is building something serious around data and AI, this is worth a look. Learn more and apply here: https://fandf.co/3XEJOrc If you do apply, tell them what you are building with confidence. The world needs more ambitious products. Thank you to MongoDB for inviting me to check this out and for sponsoring my content. #data #ai #mongodb #databases #mongodbpartner #agents #agenticai #theravitshow

Stack Overflow reinvents itself as a major enterprise AI data provider, powering smarter internal agents worldwide. 1. From Q&A Forum to Enterprise AI Infrastructure Stack Overflow is transforming into an AI data provider with Stack Overflow Internal, a secure enterprise platform designed to translate developer knowledge into structured, AI-ready intelligence for internal agents and enterprise use cases. 2. AI-Friendly Metadata and Reliability Scoring The new system exports rich metadata — timestamps, author details, tags, and coherence scores. These signals help AI models understand context, reliability, and relevance, making enterprise agents more accurate and trustworthy in technical environments. 3. Inspired by Growing Demand from Enterprises Many companies already used Stack Overflow APIs for training, pushing the platform to formalize its AI strategy. Content licensing deals with major AI labs now mirror Reddit-style agreements worth hundreds of millions in long-term revenue potential. 4. Building Knowledge Graphs for Smarter AI Agents Stack Overflow Internal uses a dynamic knowledge graph that automatically links concepts, reducing the cognitive burden on AI systems. Enterprises can apply custom tagging or rely on automated classification to strengthen internal knowledge retrieval. 5. Read–Write AI Agents That Learn as They Work Future agents will be able to write their own Stack Overflow-style questions when they detect knowledge gaps. This closes the loop — allowing AI systems to capture, document, and improve organizational expertise with minimal developer effort.

Bone AI Raises $12M to Challenge Asia’s Defense Giants With AI Robotics 1. A New Defense-Tech Disruptor Emerges Seoul–Palo Alto startup Bone AI raised $12M to take on Asia’s entrenched defense giants by building a unified AI-driven robotics platform spanning aerial, ground, and marine autonomous systems for next-generation defense missions. 2. Early Revenue and Strong Government Momentum Despite being less than a year old, Bone AI has secured a seven-figure B2G contract, generated $3M in revenue, and won a major South Korean government logistics program deploying its AI-powered UAVs and UGVs. 3. Strategic Acquisition Accelerates Growth Bone AI rapidly scaled by acquiring drone manufacturer D-Makers and its IP. The company is merging this hardware expertise with its AI models, with more strategic acquisitions planned to deepen its robotics ecosystem. 4. Building the Backbone for “Physical AI” Founder DK Lee aims to unify AI, robotics, hardware design, and manufacturing under one roof. The company wants to create the industrial infrastructure required to build intelligent machines at global scale. 5. Korea’s Opportunity in the Global AI Arms Race With Korea’s world-class manufacturing strength and rising defense investments, Bone AI positions itself as Asia’s answer to Anduril and Helsing — pushing forward sovereign AI, reindustrialization, and a modern defense robotics ecosystem.

Chinese AI Startup Moonshot AI Surpasses GPT-5 & Claude Sonnet 4.5 With Kimi K2 Thinking 1. A Breakthrough Moment for Open-Source AI Moonshot’s new Kimi K2 Thinking model has shocked the industry by outperforming GPT-5 and Claude Sonnet 4.5 on major benchmarks, signaling a major shift in the global AI power balance. 2. Dominating Key Benchmarks Globally Kimi K2 delivers 44.9% on Humanity’s Last Exam and 60.2% on BrowseComp, while leading agentic benchmarks with 93% accuracy — proving its strength in reasoning, research, and complex multi-step problem solving. 3. High Performance at a Fraction of the Cost With reported training costs around $4.6M, Kimi K2 is far cheaper than U.S. frontier models. Its efficiency-driven approach showcases China’s advantage in cost-effective AI development at large scale. 4. Technical Strengths With Some Gaps Built using a trillion-parameter MoE architecture and INT4 optimization, Kimi K2 executes 200–300 tool calls autonomously. Experts say it’s slightly behind top closed models, but the gap is shrinking rapidly. 5. A New Wave of Competitive Pressure Analysts call this a turning point for global AI. Chinese models like Moonshot, DeepSeek, and Qwen are pushing U.S. labs to innovate faster as open-source capabilities rise and pricing pressure intensifies worldwide.

ChatGPT Begins Testing Collaborative Group Chats Across Japan, New Zealand, South Korea & Taiwan OpenAI has officially launched a pilot for group chats in ChatGPT, introducing a new way for people to collaborate directly inside the app. The feature is rolling out first in Japan, New Zealand, South Korea, and Taiwan across Free, Plus, and Team plans on both mobile and web. This marks a major step in transforming ChatGPT from a one-on-one AI assistant into a more shared, social experience. OpenAI says the pilot will help them understand how users naturally collaborate in multi-person conversations. Early testers will also provide feedback that will shape how the feature expands globally. Privacy still comes first. Private chats and memory remain completely private. Group chats are invite-only, with clear controls for joining, leaving, and managing members. Extra safeguards and parental controls are applied for users under 18. How it works: Starting a group is simple - tap the people icon, add participants, or share a link. Groups can include 1–20 people. If you add someone to an existing chat, ChatGPT automatically creates a new group to preserve the original conversation. Group chats behave just like normal ChatGPT threads, but with multiple people participating. GPT-5.1 Auto provides responses, complete with search, image generation, file uploads, and dictation support. Importantly, usage limits only count when ChatGPT responds — not when humans talk to each other. Smarter social behavior: ChatGPT now knows when to step in and when to stay quiet. Users can tag “ChatGPT” to pull the AI into the discussion, and the assistant can use emoji reactions or generate personalized images using participants’ profile photos. This release follows OpenAI’s broader push into social features, including Sora 2, a TikTok-style app launched in September for sharing AI-generated videos. The group chat pilot may be small, but it signals a major evolution — ChatGPT is slowly becoming a space where AI and people interact together, not just one-on-one.