Skip to main content
AmigoFish

Kunal Kushwaha

Kunal KushwahaVideo

Format:
YouTube channel
New videos:
Irregular
Typical length:
about 10 minutes, not counting Shorts
Latest video:
Not rated yet

About this show

Kunal Kushwaha is the founder of WeMakeDevs, a global developer community of over 200,000 members focused on high-impact events and helping tech companies scale. He has previously led teams at cloud-native companies as a Field CTO and DevRel leader. Beyond community building, Kunal creates educational content on YouTube that has reached millions, and he regularly speaks at international conferences on open source, cloud native, and AI.

Videos come from the channel's public YouTube feed. The artwork is the newest video thumbnail. Last updated . Names, artwork and text belong to their publishers.

Cosmos

Rated alike

Similar by topic and creator

See the full Cosmos

Videos

  • Attackers Don't Hack In. They Log In.

    · 2 min

    Attackers don't break in anymore. They log in. Verizon's DBIR has credential abuse showing up in ~4 out of 10 breaches, and nearly half of ransomware cases last year started with a VPN login. And now we're handing AI agents access to prod through MCP servers. I hope you are auditing that... I tried this open-source tool, Pomerium, the sponsor of this post. Identity-aware proxy, every request checked against policy whether it's a human, a service, or an agent. Spun it up in a browser sandbox and killed a live SSH session mid-connection. 👉 Test it out: https://fandf.co/4xziydt Data for the stats: - https://www.verizon.com/business/resources/articles/credential-stuffing-attacks-2025-dbir-research/ - https://www.infosecurity-magazine.com/news/verizon-dbir-exploits-top-access/ - https://socfortress.medium.com/crowdstrike-2026-global-threat-report-7396c631c65d - https://thehackernews.com/2025/09/salesloft-takes-drift-offline-after.html

    Not rated yet
  • Sol, Terra, or Luna? Which GPT-5.6 Model to Actually Use

    · 1 min

    GPT-5.6 isn't one model, it's three: Sol, Terra, and Luna. Picking the wrong one wastes either your money or your time. Sol is the flagship for hard multi-step work. Terra is the free-tier default for everyday tasks. Luna is the budget pick for high-volume simple jobs, just made 80% cheaper. But here's the catch: an independent CodeRabbit benchmark found Terra burned nearly 3x more tokens than Sol on hard coding tasks, meaning the "cheap" model was sometimes the more expensive one per finished job. The rule: price the task, not the token. Sources: GPT-5.6 launch page: https://openai.com/index/gpt-5-6/ Codex models docs: https://developers.openai.com/codex/models CodeRabbit benchmark: https://www.coderabbit.ai/blog/gpt-5-6-sol-and-terra-benchmark Live pricing: https://platform.openai.com/docs/pricing

    Not rated yet
  • Deploy fast, scale smart, win a MacBook | Zerops Challenge

    · 1 hr 45 min

    🚀 Deploy fast, build with AI, and win a brand-new MacBook! Join Kunal Kushwaha and Francesco Ciulla for the official Zerops Challenge, a hands-on livestream where we'll use the Zerops Control Plane (ZCP) to build, deploy, and verify a real application with an AI coding agent. 📌 IMPORTANT LINKS 1. Sign up for Zerops: https://zerops.io/?utm_source=kunal 2. Register for the Challenge: https://www.wemakedevs.org/hackathons/zerops 3. ZCP Quickstart Guide: https://docs.zerops.io/zcp/quickstart ⚡ WHAT WE'LL DO During the livestream, we'll follow the complete ZCP Quickstart workflow: 1. Choose an AI Agent Recipe: Deploy a ready-made Zerops recipe that includes application services, managed dependencies, a ZCP workspace, Browser VS Code, and an integrated coding agent. 2. Authorize the Coding Agent: Connect your preferred agent using your own subscription or API credentials. ZCP supports multiple coding agents, including: 1. Claude Code 2. OpenAI Codex 3. Antigravity 4. Grok Build 5. Open the Remote Workspace: Work directly inside Browser VS Code with access to project files, terminal, live infrastructure, private services, and ZCP tools. 6. Ask for a Real Product Change: Give the agent a practical task in natural language, such as building a persistent task board connected to PostgreSQL. 7. Deploy & Verify the Result: Watch the agent inspect the live environment, implement the change, deploy it, test the application, and return real proof (like a working URL or verified app state). You'll see firsthand how ZCP helps an AI coding agent understand existing infrastructure, connect application services, deploy changes, and verify that the requested behavior actually works! 🎁 MACBOOK GIVEAWAY A brand-new MacBook will be given away LIVE during the event. Make sure you register, join the stream, and follow the live participation instructions for a chance to win! 📅 DATE: August 8, 2026 🔔 Set your reminder and join us live!

    Not rated yet
  • You're Prompting ChatGPT Wrong (OpenAI Says So)

    · 1 min

    OpenAI updated their prompting guide after GPT-5.6, and most of what we were all taught is now obsolete. "Think step by step," giant system prompts, role-playing as a senior engineer, their own testing found that minimal prompts actually score 10–15% better while using roughly half the tokens. The new rule: describe the result you want, not the steps. The one thing that still matters is a clear "done condition", telling the model exactly when the job is finished. Sources: - OpenAI model guidance: https://developers.openai.com - The Decoder's breakdown: https://the-decoder.com/openais-new-prompting-guide-tells-users-to-stop-overthinking-and-start-with-the-result/ - Prompting guide analysis: https://aiquill.substack.com/p/gpt-56-prompting-guide #ChatGPT #PromptEngineering

    Not rated yet
  • I’m Giving Away a $5,000 NVIDIA AI Supercomputer

    · 1 min

    I can't wait to see who will win this NVIDIA DGX Spark AI Supercomputer worth $5,000 + a functioning Iron Man helmet this month. This machine is so powerful it's like a mini data centre on your desk that you can use to run and fine-tune AI models locally. We are hosting the Into the Scrap-Verse hackathon from August 17th to 23rd, online and open to all. The theme is data scraping using the Bright Data Scraper Studio to build self-healing web scrapers. 👉 Register solo or in a team of up to 4 people: https://forms.gle/kSmhMTAkd6K9d5Cw5 Beginners and experts welcome, with live streams and networking.

    Not rated yet
  • Anyone Can Be a Web Designer Now

    · 7 min

    I hate how every AI website looks completely identical, centered headline, three generic cards, and zero personality. In this video, we test Replit Design, a spatial tool that brings visual design directly into a live code runtime. Learn how to turn prompts or Figma frames into app-ready UI, tweak layouts without wasting AI credits, and build custom websites even if you're not a designer. 👉 Try Replit Design to turn your ideas into reality: https://replit.com/refer/KunalsReplit ====================================================== Timestamps: 00:00 The AI "Sea of Sameness" Dilemma 01:17 What Replit Design Actually Is 02:42 Spatial Canvas and Design Token Governance 03:50 Iteration Made Simple 05:06 Insights for Smooth Development and Resource Efficiency 06:27 The Takeaway for Designers and Builders #Replit #WebDesign #vibecoding

    Not rated yet
  • Live MacBook Giveaway this Saturday

    · 58 sec

    This Saturday I am giving away a MacBook live on stream with Francesco + prizes worth $5,000. Join us for the Zerops Challenge, a hands-on livestream where we'll use the Zerops Control Plane (ZCP) to build, deploy, and verify a real application with an AI coding agent. 👉 Register now: https://www.wemakedevs.org/hackathons/zerops

    Not rated yet
  • Is Cursor Composer 2.5 Actually Worth the Hype?

    · 1 min

    Try it out: https://cursor.com/composer

    Not rated yet
  • I quit my job and hired an AI co-founder

    · 22 min

    AI tools make it easy to spin up a prototype in minutes, but 42% of startups still collapse because there’s no market demand. Building fast doesn't matter if you’re vibe coding blind without validating your idea first. I started testing "Vibe Solutioning" using Rocket 1.0 to fix this. It acts like an AI co-founder, running deep research and tracking competitors in real time *before* you prompt. That research then flows directly into production-grade Next.js or Flutter code so you actually build the right thing. 👉 Try out the world's first Vibe Solutioning tool: https://www.rocket.new/?utm_source=youtube&utm_medium=kunal&utm_campaign=rocket1o =========================================================== Timestamps: 00:00 Why 42% of new AI startups collapse 01:10 The world's first "Vibe Solutioning" platform 02:12 Problem with AI research tools 03:34 Competitor tracking 04:55 Vibe-coded websites look the same 06:03 Demo 20:49 Closing remarks #vibecoding #startups

    Not rated yet
  • Can AI Agents Actually Run a Business?

    · 1 min

    Are AI agents finally ready to run a real business, or is it all just hype? In this video, we put @Accio_official, a native desktop AI agent developed by Alibaba.com, to the test on a real-world e-commerce sourcing campaign. Instead of just giving generic advice like a standard chatbot, Accio Work actually executes tasks directly on your computer: navigating suppliers, pulling pricing into spreadsheets, drafting outreach messages, and setting up automated workflows. We also cover the essential security side, how its local-first architecture and human-in-the-loop permissions keep your sensitive business data and accounts completely safe. #AccioWork #AgenticBusiness #AccioAgent #MyAcciowork #AIAgent #BusinessAutomation

    Not rated yet

More in Education

See all Education shows