How to AI

BREAKING: Andrew Ng, founding Lead of Google Brain, just launched his new course: "Agentic AI". Here’s what you get: 1. Learn to build AI agents that think and act. 2. Master four key agentic design patterns:   • Reflection: Teach your agent to review its own work and improve.   • Tool use: Let your agent pick the right tool—search, email, code, and more.   • Planning: Break big tasks into small steps, and get things done.   • Multi-agent collaboration: Build teams of agents, each with a job, working together. 3. Get hands-on with real projects and code. 4. Learn from Andrew Ng, the mind behind Google Brain. To access it: https://lnkd.in/dyHD9PRr Start here if you’re new: how-to-ai.store

Uncomfortable Truth for Consultants: AI is taking your job, & it’s not even trying hard. You need 1 AI tool to do what half the industry does. ✦ Gamma to make slides, better and faster. ☑ Go to Gamma.app. ☑ Start a new slide with AI, from scratch. ☑ Or upload your existing PPT directly there. ☑ Let Gamma remake your slides with templates. Gamma does what juniors do. Only faster. It writes, designs, and formats. It even adds charts and images. You get a full slides in minutes. Deloitte’s AI report was full of errors. They had to refund the government.  The message is clear: AI is already replacing consultants, but not all AI is equal. If 70% of the work is slides, and AI can do most of it, half the junior consultants are gone. Gamma has 50 million users. That’s not hype. That’s a signal. AI will automate 30% of consulting tasks.  The first to go: data analysis and report prep.  The junior work. The grunt work. The future of consulting is not PowerPoint.  It’s not even ChatGPT.  It’s Gamma.

This is the single best Sora 2 video I've seen. This video is 100% AI-generated. Prompt: "body cam footage of a police officer pulling over a car with a bunch of sliced ham as the driver. After being questioned the ham drives away with the police officer trying to chase on foot (Shaky bodycam) before the video cuts off. Every frame is made by AI. No actors. No real cars. Just a prompt and Sora 2.

Anthropic quietly dropped a Context Engineering guide. Save these 10 rules to master it: To access the full guide, go to: https://lnkd.in/dMU8mTar. 1. Treat context as a finite resource   ↳ The context window is limited.   ↳ More tokens can mean worse results.   ↳ Only include the most relevant, high-signal data. 2. Go beyond prompt engineering   ↳ Don’t just write static prompts.   ↳ Manage the whole context: system prompts, tools, message history, and runtime data. 3. Make system prompts clear and minimal   ↳ Avoid vague or brittle instructions.   ↳ Use structured formats like Markdown or XML.   ↳ Focus on signal, not just brevity. 4. Build tools for efficient agent behavior   ↳ Each tool should do one thing well.   ↳ Outputs must be compact and unambiguous.   ↳ Keep tool functions separate and clear. 5. Use canonical, diverse examples   ↳ Pick a few high-quality, representative cases.   ↳ Skip rare edge cases.   ↳ Show the model what “good” looks like. 6. Support just-in-time context retrieval   ↳ Let agents pull in data as needed, like human memory.   ↳ Use lightweight references—file paths, queries, or links—instead of loading everything up front. 7. Apply a hybrid retrieval strategy   ↳ Pre-load key data for speed.   ↳ Let agents explore and fetch more as needed.   ↳ Balance speed and flexibility. 8. Enable long-horizon agent behavior   ↳ Support agents working over hours, days, or sessions.   ↳ Summarize old context to save space.   ↳ Use structured notes and sub-agents for complex tasks. 9. Design for progressive disclosure   ↳ Let agents discover info step by step.   ↳ Use tools like directory browsing to reveal context as needed.   ↳ Context should emerge and refine through use. 10. Curate context dynamically and iteratively   ↳ Context engineering is never “done.”   ↳ Use feedback to refine what’s included and how it’s formatted.   ↳ Keep improving with every run.

I can't believe this anime clip is 100% Sora 2. Here’s exactly what’s new: OpenAI’s Sora 2 is the ChatGPT moment for video.   Now, it even generates sound and dialogue. 1. Sora 2 creates videos that look and move like real life.   2. It handles tough stuff: gymnastics, water, even objects that bounce or break.     3. Sora 2 now adds voices, sound effects, and music that match the video.   4. You can give it detailed, multi-shot instructions. It keeps the world and characters consistent, even across anime, movies, or real scenes.   5. Upload a short clip of yourself—Sora 2 puts your face and voice into any scene.   6. The new Sora app works like TikTok: scroll, remix, add cameos, and share.   7. Safety is built in: parental controls, teen protections, and strong rules for using anyone’s likeness.   8. Free to try in the US and Canada. Pro version for ChatGPT Pro users. API is coming soon. This is the start of AI that can simulate whole worlds.  Sora 2 is the next step for AI and video. Credit: @naegiko on X. If you want more of these, repost ♻️ this video.

HUGE: OpenAI dropped Sora 2 out of nowhere. This is how wild Sora 2’s physics are. Quick heads-up on Sora 2 (vs. Sora 1): ☑ Sora 2 is the new king of AI video. ☑ Sora 2 nails physics, realism, and control. ☑ Sora 1 was... clumsy. In less than a year, Sora went from: "That’s a laughable AI video." to "Did that gymnast just stick the landing?" ♻️ Repost this video to help others get up to date. To master AI with English (not code), go to → how-to-ai.store.

BREAKING: Google released a professional headshot prompting guide. 1. Take a clear photo of yourself. 2. Upload it on Gemini. 3. Copy & paste the following prompt: 4. I wrote a (free) nano banana prompting guide. 5. Go to this blog: https://lnkd.in/d7xW8Qze ____ Prompt: "A professional, high-resolution, profile photo, maintaining the exact facial structure, identity, and key features of the person in the input image. The subject is framed from the chest up, with ample headroom and negative space above their head, ensuring the top of their head is not cropped. The person looks directly at the camera, and the subject's body is also directly facing the camera. They are styled for a professional photo studio shoot, wearing a smart casual blazer. The background is a solid '# 141414' neutral studio. Shot from a high angle with bright and airy soft, diffused studio lighting, gently illuminating the face and creating a subtle catchlight in the eyes, conveying a sense of clarity. Captured on an 85mm f/1.8 lens with a shallow depth of field, exquisite focus on the eyes, and beautiful, soft bokeh. Observe crisp detail on the fabric texture of the blazer, individual strands of hair, and natural, realistic skin texture. The atmosphere exudes confidence, professionalism, and approachability. Clean and bright cinematic color grading with subtle warmth and balanced tones, ensuring a polished and contemporary feel." tl;dr: ✦ Lighting is everything. ✦ Get your angle right. ✦ Dress for the job you want. ✦ Use a solid, neutral studio background. Help others create better headshots by ♻️ reposting this. PS: Ruben writes a newsletter to 129,000+ readers. Join us for free at how-to-ai.store.