AI-fokus 2025
| title | author | length |
|---|---|---|
| Everything You Need to Know About Running LLMs Locally | Cedric Clyburn | 45m |
Explains how to run open‑source LLMs locally with concrete stacks: demos use Llama C++ and Olama for local serving, quantized Llama 3.1 at 4‑bit to cut memory… | ||
| GPT-4 vs Starcraft II – Strategic Decision Making using Large Language Models | Alan Smith | 44m |
Demo shows GPT-4 controlling StarCraft II through a Python API, producing 20-step build orders and switching between defend or attack strategies, with attack… | ||
| 15 Months of building AI agents in 15 minutes | Victor Dantas | 14m |
Builds on 15 months of creating AI agents for Google Cloud customers and claims practical rules that work in production: use critical user journeys to tie… | ||
| AI Agents for the Rest of Us | Korey Stegared-Pace | 46m |
Shows concrete code using Semantic Kernel and Azure AI Agent Service to build multiagent workflows (a medical demo with about five to six agents: patient… | ||
| Mastering the Model Context Protocol | Carl-William Ersgård | 35m |
Demonstrations build MCP servers with the official TypeScript SDK, expose tools that call ClickUp and GitLab APIs, and use standard-input/stdout or HTTP… | ||
| MCP for LLM Agents, Infrastructure APIs & Data Services | Michael Hunger | 46m |
Explains how the Model Context Protocol makes tool calling reusable by wrapping REST/DB APIs as MCP servers and connecting them to clients like Cloud Desktop… | ||
| Navigating AI Native Development | Simon Maple | 53m |
Argues software shifts from code-centric to spec-centric development, demonstrates Bolt.new, Cursor and Copilot generating working UIs (a Bolt app shows 242… | ||
| Agentic AI and the Future of AI | Heli Helskyaho, Matias Helskyaho | 42m |
Claims that 2025 is the year of agentic AI and demonstrates a working car‑repair and travel agent built in Python using LangChain, LangGraph and LangSmith… | ||
| My Bank Meets AI: Smarter Decisions Beyond Fancier Spreadsheets | César Soto Valero | 16m |
Generative AI reaches near human level by 2025 and banks move from prototypes to production with clear wins and clear limits. Examples include a LLM trained on… | ||
| Using AI Across the Dev Lifecycle | Mattias Soderberg | 18m |
Survey of ~5,000 devs shows 78% use AI somewhere and claims AI speeds only ~21% of a developer day, creating a downstream bottleneck; proposes platform-level… | ||
| Apropos DeepSeek | Dan Bergh Johnsson | 44m |
DeepSeek releases an open model (R1 under MIT license) that claims real gains: a 671 billion-parameter base with 37 expert activations and training on “two… | ||
| RoboCoders: Judgment Day – AI IDEs Face Off | Baruch Sadogursky, Viktor Gamov | 50m |
AI IDEs are asked to build a webcam-driven smart light bulb controller from scratch and mostly deliver: Windsurf, a Copilot-powered VS Code agent and JetBrains… | ||
| The AgentVerse - Build fast, Scale smarter | Begum Firdousi Abbas | 43m |
Demonstrates building Amazon Bedrock agents that use Claude 3.5 Haiku or Amazon Nova, embeddings stored in vector stores (Titan or pgvector) and Lambda-backed… | ||
| Trustworthy AI: The Key to Unlocking High-Stakes Applications | Johan Falk | 38m |
Argues that trustworthy AI is the enabler for high‑stakes use and cites concrete failures: a Dutch benefits‑fraud system that wrongly flagged migrants and… | ||
| AI Implementation: Navigating the Enterprise Lifecycle for Technical Leaders | Jesper Hagberg | 24m |
Maps the enterprise AI lifecycle from idea intake to production and argues leaders must choose between buy from existing vendor, buy a new specialist, build… | ||
| Productivity Hacks from Babylon to Sweden | Ahmad Mudhar | 15m |
Starts from childhood in Babylon and moves to concrete Swedish examples of AI replacing tedious tasks: a property portal uses AI to auto‑categorize maintenance… | ||
| Trusting AI to Modernize Software at Scale | Tim te Beek | 45m |
Moderne builds a mass-scale refactoring pipeline on OpenRewrite and experiments with GPT-4o for recipe generation but finds raw LLM outputs often fail to… | ||
| How is AI reshaping tech teams | Paula Bruneheim | 12m |
Stanford collaboration analyzes 50,000+ contributors and finds top 25% produce six times more output than bottom 25%, with AI widening that gap while junior… | ||
| Bringing stories to life with AI, data streaming and generative agents | Olena Kutsenko | 44m |
Prototype builds a mini Stardew Valley village where characters walk, start dialogue when within ~1,024 pixels, and exchange memories; position events stream… | ||
| Accelerated time-to-value in business transformation powered by genAI | Juho Nevalainen | 28m |
Company claims a repeatable method to cut years into weeks by tying business strategy to concrete user journeys, showing a live demo that generates AI use… | ||
| AI Do’s and Don’ts for Managers: Bridging Strategy and Technology | Ulf Larsson | 43m |
SEB shares concrete practice: a separate SEBX lab, a pilot “use‑case competition” that floods with ideas, and an incident where an AI credit model produced… | ||
| How Do We Handle AI Opponents? | Jonas Hermansson | 11m |
Argument: resistance comes from fear, not failure; examples: an AI-generated portrait with six fingers from shutt, buggy code and two new errors from chat GPT… | ||
| AI fokus 2025 intro | 01m | |
AI-fokus 2025 opening video | ||
| Data-Empowered AI Assistants | Michael Hunger | 28m |
Claims that consolidating metadata into a Neo4j knowledge graph gives AI assistants contextual access rather than fragmentary vector hits, combines node… | ||
| nothing matches | ||