Crazy how the past year has changed things for me. Hi, I'm Lipsa 👋 Started @Dell in 2019 as an automation developer. Built some Python scripts. Automated boring stuff using RPA. Nothing crazy. Thought it was pretty cool... but never imagined AI would completely take over my timeline (and my career). Went from the classic 'AI sucks at writing' skeptic to using AI for getting unstuck. And now, I: > Write AI pieces for billion-dollar brands like @Hubspot > Build out AI and automation workflows for my Web3 agency @bbtcontent > Handle socials & marketing for an exciting agentic AI startup. I want to share everything I've learned (and am learning). Excited to begin this AI arc with y'all. What was your "aha" moment with AI?

Find all videos here: piped.video/Ub3GoFaUcds piped.video/yT84Y5zCnaA piped.video/Q5baLehv5So If you found this useful, follow @itslipsadas for more AI breakdowns.

Stanford just dropped 5+ hours of free lectures on LLM architecture. Covers everything from: - Tokenization & embeddings - Attention optimizations - Prompting & Chain-of-Thought - Inference engineering - Model families and so much more. Link below ↓

Not only SaaS, but retail is also changing with AI. GAP is using Google Cloud Revolve’s got an AI stylist The Body Shop has a predictive supply chain 85% of retail executives already have AI capabilities in place. Retail's next decade won't have stores, it'll be AI shopping sidekicks.

Google is building its own version of OpenAI's AI agent. Soon you'll be able to let it browse, make bookings, and complete repetitive tasks for you. It's said to use Gemini 2.5 Computer Use model.

This simple prompt hack instantly reveals blind spots and gives 10x legit responses. “On a scale of 1–10, how confident are you? If not 10, then why?”

This poster shows every notable humanoid robot from 2023 vs 2025. We went from 13 robots (mostly demos) to 30+ (some already working in factories). Tesla, BMW, Mercedes are all testing humanoids for real tasks. It’s happening.

Gemini 2.5 Deep Think just achieved SOTA on FrontierMath. - Got 29% of the advanced math problems. - 10% on Tier 4 (hardest) Gemini 3 is going to be insane.

82% of ChatGPT users have never tried other GenAI tools. yes, you're still early.

7 prompting lessons that will instantly improve your results (not hacks or complicated roleplays) > swap long instructions for few-shot examples. > version your prompts systematically, use tools like Promptfoo or Vellum. > build proper test suites to know what works. > get domain experts involved in writing and testing prompts. > start simple before trying complex techniques. > different models need different prompts, adapt and revise. > start small and let AI guide improvements.

OpenAI’s going all in on AI agents. Their rumored Agent Builder will let you build agentic workflows with simple drag and drop, like: → lead routing → internal research bots → automated support flows It probably won’t replace n8n, as that’s more of a visual coding tool than an agent builder. But massively lowers the barrier to production-grade agent workflows for everyone else.

Use this prompt to find automations that save you 10+ hours/week

Claude Sonnet 4.5 is now ranking #1 on LMArena (tied with Gemini 2.5 Pro & Opus 4.1). Insane run for Gemini to lead six months straight.

I don’t know who’s actually trying 150 AI tools from those lists. Most scroll, maybe bookmark, and never open again. So here’s just 10 tools worth keeping. (yes, helps finish hours of work in minutes) 1. Gamma for presentation 2. Claude for writing/coding 3. ChatGPT for everyday stuff 4. Wisperflow for AI voice dictation 5. Perplexity Comet for agentic browsing 6. Fathom for meeting notetaker 7. Notion for task management 8. Krea AI for video/image gen 9. Canva AI for design 10. SuperX for twitter

Still can’t believe how many are sleeping on Google AI Studio. You get free Gemini new model access, multimodal processing, and you can prototype workflows without burning API credits. 100% free. One of the most underrated AI tools by far.

How to 10x your AI fluency (without getting overwhelmed): 1. Set up Perplexity Comet for agentic actions (quick research, daily updates). 2. Create Gemini Gems for custom content assistants 3. Connect Claude to your CRM/Notion with MCP servers 4. Start a personal prompt library (refine & reuse as you go) 5. Make to build for mini automations (scanning trends, auto-generate reports). 6. Run competitor analysis or market research with Grok4 7. Combine Perplexity + Gamma to research and build slide decks. 8. Use Notion AI agents to automate content & brainstorm. 9. Create dedicated Claude projects with brand guidelines for ongoing work. 10. Vibecode landing page, portfolio site with bolt or cursor

- Open Perplexity - Go to Personalization Settings - Save this system prompt Now ask anything to a personal research analyst & get structured breakdowns with a confidence score.

This 5-stage AI workflow makes job hunting 10x smarter and faster: → analyze successful resumes in your target role → understand industry pain points → research company strategy & challenges → prepare specific examples for their problems → plan automation workflows Basically, everything you need to stand out from 1k other applicants.