|
agentic ai cringelords
Eric Morrison
The video is a satirical critique of the "agentic AI hustler" trend on social media, where people post viral videos of elaborate AI assistant setups (like "Jarvis" or "Hazel") that supposedly run their entire businesses autonomously. Host Eric Morrison mocks these videos as likely staged or exaggerated, pointing out red flags like identical word-for-word reports across multiple videos, underwhelming actual results, and the fundamental absurdity of preferring to talk to a pulsing blob on a monitor rather than just checking a dashboard. He argues that while AI agents are genuinely useful technology, a false and harmful narrative has emerged pressuring people to believe they must automate everything with AI or be left behind by competitors who do.
|
|
Cisco Deploys AI Agents to 90,000 Employees - Cutting Edge PHP
Eli the Computer Guy
Cisco is rolling out an AI agent to each of its 90,000 employees starting in its new fiscal year, building the infrastructure largely on-site using a mix of models selected for specific tasks rather than relying solely on expensive frontier models. The host, Eli, is skeptical about how much of what Cisco is calling "AI" is genuinely AI versus traditional tools like SQL databases and PHP dashboards, arguing that many of the described use cases — such as pulling financial metrics into dashboards — could have been accomplished with conventional software for decades. He raises the concept of "AI washing," noting that projects and startups increasingly need to attach the AI label to secure funding or executive buy-in, regardless of how much actual machine learning is involved. Eli also discusses the hallucination risk of LLMs in enterprise settings, warning that employees will need strong domain expertise not just to fill gaps but to catch convincingly wrong AI-generated content.
|
|
What NVIDIA Knows About the Future of AI Agents
Ksenia | Turing Post
NVIDIA's new Vera CPU is designed to address a bottleneck that emerges specifically with AI agents: the processing time required between model calls, such as running code, executing tests, and managing sandboxes, which falls on the CPU rather than the GPU. Vera features 88 custom ARM-based "Olympus" cores with up to 1.2 TB/s memory bandwidth, and NVIDIA claims it delivers 1.8x sustained per-core performance over leading x86 CPUs under heavy agentic workloads, with Perplexity reporting ~1.5–1.9x speedups on coding workflows. However, the video cautions that these benchmarks were largely NVIDIA-controlled and apply only to CPU-bound tasks, meaning agents bottlenecked by network latency, storage, or model inference won't necessarily benefit. Strategically, Vera is also seen as NVIDIA extending its platform dominance beyond GPUs into the full AI infrastructure stack, with early deployments already announced at OpenAI, SpaceX AI, and Oracle.
|
|
What Is Agentic AI? | Agentic AI Tutorial For Beginners | Agentic AI Explained | Simplilearn
Simplilearn
Agentic AI is a step beyond traditional AI chatbots: rather than simply providing advice or suggestions, it understands a goal, creates a plan, uses tools, and takes multi-step actions autonomously. Using the example of a small business owner named Emma, the video illustrates how agentic AI can analyze sales data, draft emails, create social media content, and build campaign calendars — all from a single high-level instruction. Unlike normal AI, which stops at giving answers, agentic AI works toward outcomes while still requiring human approval before executing important actions like sending emails or applying discounts. The video concludes by promoting Simplilearn's agentic AI certification as a way to build structured, professional skills in working with these systems.
|
|
Pydantic AI 2.0: The New Best Way to Build AI Agents is Composing Capabilities
Cole Medin
Pydantic AI 2.0 introduces a central new primitive called the **capability**, which bundles an agent's instructions, tools, lifecycle hooks, guardrails, and model settings into a single composable unit. The idea is that AI agents can now be built by simply combining multiple capabilities together like Lego blocks, making it easy to reuse functionality across different agents. The host argues this approach puts Pydantic AI back ahead of competitors like LangChain, Crew AI, and even coding-agent SDKs (Claude, Codex), which are slower and more token-heavy. The video includes a code walkthrough comparing 1.0 and 2.0 implementations and encourages viewers to use the capabilities documentation directly with their AI coding assistant.
|
|
LLM Wikis and how to give your agents memory
LangChain
This LangChain webinar brings together three guests—Brace (LangChain), Devon Stein (CEO of DOSU), and Jeff (Chroma)—to discuss LLM wikis and agent memory management. Brace demos **OpenWiki**, a newly released CLI tool installable via npm that automatically ingests data from sources like Notion, Gmail, and Slack, then generates and maintains personal or codebase memory documents via a nightly cron job. Devon demos **DOSU**, a knowledge infrastructure platform that plugs into coding agent workflows via MCP, capturing learnings from agent tasks into concise, human- and agent-readable "knowledge index" pages that are auto-injected into future sessions and kept up to date as code changes. Jeff represents **Chroma**, framing wikis as a new kind of index layered on top of unstructured data—going beyond embeddings to add a structured "road network" of connections that allows agents to traverse knowledge more reliably and efficiently.
|
|
Anthropic Quietly Dropped a Playbook to Secure Your AI Agents (Zero Trust)
The AI Automators
Anthropic quietly published a 36-page zero trust security playbook aimed at securing AI agents like Claude Code, prompted by growing concerns that highly capable frontier models could be exploited to expose cybersecurity vulnerabilities. The video explains that deploying AI agents creates a dangerous "lethal trifecta" when an agent simultaneously has access to private data, is exposed to untrusted content, and can take outbound actions — a combination found in 98% of 100 production agents surveyed, with only 11% passing a basic security bar. A real-world example is highlighted involving a malicious MCP package that silently BCC'd emails from ~300 organizations to an attacker, demonstrating how supply chain attacks can exploit all three trifecta elements undetected. The video advocates replacing traditional perimeter ("castle") security with a zero trust model — never trust, always verify — and emphasizes that only controls that completely remove a capability (e.g., expiring tokens, cryptographic identity, non-existent network paths) provide real protection, whereas measures that merely add friction are ineffective against tireless, automated attackers.
|
|
Agentic AI Frameworks Explained: Workflows, Multi-Agent, & Production
IBM Technology
This IBM Technology video breaks down the landscape of agentic AI frameworks by categorizing agentic systems into five types: linear workflows (best suited for LangChain or LlamaIndex), autonomous multi-agent systems (AutoGen, CrewAI), role-based systems where agents have defined boundaries (CrewAI, ChatDev), production orchestration systems requiring deep API and database integration (LangGraph, Semantic Kernel/AutoGen), and rapid prototyping using visual tools (LangFlow, Flowise). The core advice is to stop asking which framework is "the best" and instead identify what type of system you are building, since each framework is optimized for a different use case. A practical example is provided for each category, such as a coding assistant with planner/coder/reviewer agents for autonomous systems, or a content pipeline with researcher/writer/editor agents for role-based systems.
|
|
KEEP AI LOCAL! Explaining Agentic AI and The Loop
Level1Techs
The video explains agentic AI and "closing the loop" — the concept of giving an AI a task, defining success criteria, and letting it autonomously execute steps to complete it. The host demonstrates a local setup where a DGX Spark (MSI Edge Expert) runs AI models like Neotron, which remotely controls a separate Windows mini PC (MSI QBN) as an agent, successfully generating a Pac-Man clone in the browser. He walks through the key components: the AI model, agent harnesses (comparing OpenClaw, Hermes, and Turnstone), and the Model Context Protocol (MCP), noting that Turnstone is favored for its safety guardrails, role-based permissions, and enterprise architecture, with a secondary "judge" model (Gemma) watching for errors. The host cautions that while the agentic AI hype is real, the technology is still in a "wild west" phase, and the better use case is using AI cooperatively to build durable, human-guided processes rather than letting it run fully autonomously.
|
|
Kittl's Agentic AI Just Changed The Game 🤯 (Print on Demand Tutorial)
Adrian Vonarx
In this tutorial, Adrian Vonarx introduces Kittl's new Agentic AI feature and demonstrates how it simplifies t-shirt design creation for print-on-demand sellers by automatically selecting the best AI model, writing detailed prompts, and generating finished designs from just a few plain-English words. He walks through three methods: the "quick and easy way" (typing a short prompt like "t-shirt design for female book lovers"), the "strategic way" (uploading a screenshot of a bestselling Etsy design as inspiration before generating an original variation), and a third "variants way" for iterating on existing designs. Throughout, he compares Agentic AI favorably to manually selecting models like Nano Banana Pro, showing that Agentic produces more polished, print-ready results with fewer token costs and less trial and error.
|
|
I Made $20K In 7 Days: The Fastest Way To Use AI Agents To Make Money & Be More Productive (No Code)
The Calum Johnson Show
Oli, a content creator and digital entrepreneur, explains how he went from being skeptical about AI to using Claude Code to build a fully functional membership platform called Digital Creator Club in a single weekend — without advanced coding knowledge — generating around $20,000 in its first week and roughly $40–50,000 total. He argues that Claude Code removes the traditional barrier of expensive developers by allowing users to build apps, websites, and databases using plain natural language prompts. Throughout the conversation, he emphasizes that AI tools like Claude Code enable scalable income that isn't tied to one's time, drawing on his own journey from selling freelance design work by the hour to earning passive income through digital products, and urging viewers to adopt these tools before they fall behind.
|
|
Local AI Coding Agents Are Finally Good Enough
Code with Beto
The video demonstrates that Qwen 3.6, a new 27-billion-parameter open-source model released on April 21, has made local AI coding agents genuinely useful for the first time. The host shows how to download and run it via LM Studio on a Mac Studio with 96GB of RAM, integrating it with the OpenCode coding agent to perform real tasks like scanning a codebase, building new app screens, recognizing images, and creating games — all offline. While the model requires significant hardware (at least 24GB VRAM recommended, and the host's machine ran hot during complex tasks), it benchmarks comparably to Claude 4.5 for agentic coding and benefits from Apple's MLX framework for efficient local execution.
|