What's new?

Generative AI explained from the inside. Follow one sentence from text to tokens to vectors and back out, then see how foundation models, embeddings and AI agents work, with no maths. ⏱ Chapters 01:25 What makes a model generative 03:20 Tokens and chunking 05:30 Embeddings, vectors and semantic distance 09:50 Vector databases on AWS 15:30 The foundation model lifecycle 17:00 Agentic AI: tools, memory, MCP, multi-agent patterns 20:10 The advantages of generative AI 21:10 The hard limits: hallucination, nondeterminism, interpretability 24:40 The AWS generative AI stack: Bedrock, JumpStart, Amazon Q, AgentCore 27:40 Recap: twelve things 📘 Free resources · Ep 1 – AI, ML and deep learning fundamentals: • AI Fundamentals Ep 1 What AI Actually Is - https://www.youtube.com/watch?v=rD97J5I1DQY Ep 2 – Use cases and the ML lifecycle on AWS - https://www.youtube.com/watch?v=2rwLAaAcMLY&list=PLJVLmAw4N6yw · Full playlist: AI Fundamentals (AWS AI Practitioner AIF-C01) → https://www.youtube.com/watch?v=2rwLAaAcMLY&list=PLJVLmAw4N6yw 🔗 Next in the series → Episode 4: Making models work: prompting, RAG, fine-tuning and evaluation This episode explains what a foundation model is, how it is pre-trained on unlabeled data, and how tokens and chunking shape both cost and accuracy. You get embeddings explained through one worked example, the vector database options on AWS, and transformers, diffusion models, GANs and VAEs by how to recognise each one. It also covers agentic AI with tools, memory and Model Context Protocol, the hard limits of hallucination and nondeterminism, and token-based pricing. It finishes with what Amazon Bedrock is, where SageMaker JumpStart and Bedrock AgentCore fit, and how to choose a model. #AWSAIPractitioner #GenerativeAI #AIforBeginners My Linkedin: https://www.linkedin.com/in/jacinthp/ --------------------------------------------- Github: https://github.com/jacinthpaul --------------------------------------------- Website: https://www.programstrategyhq.com/ ---------------------------------------------

Claude Code Funny

I'm taking part in GrowthX®'s Build Sprint S4. I built this with AI as the warm-up. Starting tomorrow, I'm building something around Wealth Management. Two weeks, in public. Follow along and my DMs are open. #GrowthX #BuildSprint

AI Fundamentals Ep 2 | ML Use Cases & the ML Lifecycle | AWS AI Practitioner AIF C01 Most AI projects fail at problem framing, not at the model. This episode of the AWS Certified AI Practitioner (AIF-C01) series covers Domain 1, how to tell when rules beat machine learning, how to pick the right technique, and the full seven-stage ML lifecycle on AWS. What you'll learn: • When a problem needs ML, and when simple rules are the better answer • Classification vs regression vs clustering vs dimensionality reduction, plus a pick-the-technique drill • Real-world AI use cases: computer vision, NLP, intelligent document processing (IDP), fraud detection, forecasting, recommendation systems, knowledge bases and agentic AI • The seven-stage ML lifecycle, from business goal to retraining • Traditional ML vs foundation models, and how to choose between them • The AWS AI/ML stack: AI services, Amazon SageMaker AI, and infrastructure • A SageMaker AI tour: Data Wrangler, Feature Store, JumpStart, Canvas, Experiments, Model Monitor • Deployment options: real-time, batch transform, asynchronous and serverless inference • Model metrics (accuracy, precision, recall, F1, MSE, R squared) and business metrics • MLOps fundamentals • Three exam-style practice questions with full answer explanations AI Fundamentals series Ep 1 – AI, ML and deep learning fundamentals: https://youtu.be/rD97J5I1DQY Ep 2 – Use cases and the ML lifecycle on AWS (this video) Ep 3 – Generative AI fundamentals: tokens, embeddings, foundation models (coming next) Program Strategy HQ covers AI, delivery and programme strategy for people who run real work. Subscribe for the rest of the series. #AWSCertified #AIF-C01 #AWSAIPractitioner #MachineLearning #AmazonSageMaker My Linkedin: https://www.linkedin.com/in/jacinthp/ --------------------------------------------- Github: https://github.com/jacinthpaul --------------------------------------------- Website: https://www.programstrategyhq.com/ ---------------------------------------------

Jev is the first System One model from TypeSafe AI. Instead of generating text, it returns typed decisions with probabilities, and answers all your questions in one parallel pass. Key facts (published by TypeSafe AI): • 40–200× faster than frontier LLMs on comparable tasks • 70–500 ms per call, end to end • $0.042 per 1M input tokens, output is free • Answers only from the options you define, so there are no schema violations • Trained with RLCD (Reinforcement Learning for Calibrated Decisions) Use cases: routing, classification, scoring, guardrails for LLM output, controlling agents, and map-reduce over large datasets. Not built for chat, code generation, or written explanations. Pair it with an LLM in your pipeline. Early access launched September 15, 2026. Learn more: typesafe.ai #AI #Jev #TypeSafeAI #MachineLearning #LLM #AIAgents #ArtificialIntelligence #TechExplained #Shorts

At the UN General Assembly on September 22, President Trump said "artificial intelligence" will now be called "super intelligence" in all U.S. government documents, and he hopes the rest of the world's too. His reason: "artificial" sounds fake. "It's not fake. It's actually amazing, but we have to be careful." Before the speech, he asked followers on X and Truth Social to vote on a new name. The options were Superior, Extreme and Supreme Intelligence. He went with one that wasn't on the list. "Whoever wins super intelligence wins." What would you call it? Tell me in the comments 👇 #AI #SuperIntelligence #Trump #UNGA #ArtificialIntelligence #TechNews #Politics #News #Shorts