Why You SHOULD Piss Off Your AI (You're Using It Wrong!) скачать в хорошем качестве

Why You SHOULD Piss Off Your AI (You're Using It Wrong!)

📝 Get 3x more interviews with https://resumeflow.app 🥅 Accomplish your goals with our coaching portal at https://thrive.techcoachralph.dev 🏖 Exclusive Vibe Code & Chill at https://techcoachralph.dev I used to accept polite, generic AI outputs — until I learned to piss the model off. This video walks through Tom B's brutal, three-step workflow that forces LLMs to critique and rewrite their own work so you get concrete, testable designs instead of fluff. What you'll see and learn: The core thesis: if your prompts never 'piss off' the AI, you're leaving value on the table. Push it to find flaws. The three-step workflow (fast, repeatable): 1) Get the first draft — let the model produce without interruption. 2) Ask the four-word critique: 'Why does this suck?' — force specific, brutal feedback. 3) Rewrite — fix every problem the model found and push again until the output is strong. Live demonstration highlights: How an initial generic QA prompt suggested agentic workflows, then collapsed into vague ideas. After the critique, the model surfaced real problems (orchestration tax, maintenance trap, hallucination/brittleness). The rewrite produced concrete architectures: context-injection agents, blocker-detective monitors, log-translators, self-healing test selection, and an aggressive shadow QA agent. Risks and trade-offs discussed: Hallucinated bugs and wasted investigation time Cost explosion from running large models at scale Black-box accountability: humans still own production bugs Why conservative, deterministic agents are safer for some tasks Who this is for: QA engineers transitioning to reliability architects Prompt engineers and LLM practitioners Engineering leaders designing safe agentic workflows Actionable takeaways: Force critique: use 'Why does this suck?' to break first-draft complacency. Start conservative (summaries, log translators) and only scale autonomy after validating safety and cost. Shift skills: think orchestration, observability, and human-in-the-loop safety. Try the prompt pattern and see how much stronger your AI outputs become. More resources and example prompts at techco.dev and tickle.dev. AI PromptEngineering QAtoReliability AgenticAI LLM PromptDesign ReliabilityEngineering AIAgents AutonomousAgents Testing DevOps SiteReliability #AIprompts #promptengineering #Whydoesthissuck #TomBprompts #LLMcritiqueworkflow #agenticAI #QAtoreliability #reliabilityarchitect #autonomousagents #selfhealingpipelines #shadowQAagent #logtranslator Timestamps 0:00 - Intro: Discussing Tom B's "piss off your AI" thesis 2:58 - Explaining how to get the AI to care (why it matters) 5:54 - Prompting Gemini with QA engineer day-to-day activities 8:50 - Discussing why this stage of AI evolution feels generic 11:46 - Explaining the context-injection fix for prompts 14:44 - Describing an agent component that identifies changes 17:40 - Demonstrating the coaching portal and its features 20:34 - Outro: Final thoughts and sign-off Join this channel to get access to perks:    / @techcoachralph   Please like, share, and subscribe to the channel. Leave a comment and we’ll be sure to respond. Become a Tech Barbarian! Join the Tech Coach Ralph Patreon:   / techcoachralph   Technical Coaching with Ralph: https://www.techcoachingwithralph.com X: https://www.x.com/techcoachralph Instagram:   / techcoachralph   Sign up for Thoughtless Automatic Investing with Acorns: https://share.acorns.com/rgehy84?advo... Stash your Money in the right places: https://get.stash.com/ralph_2tsv8q3 #softwareengineer #softwaredevelopment #qaengineering #qaengineer #softwaretesting #python #java #programming #coding #technicalcoaching