Hey, I’m Pranay. I’ve been in tech for 20 years, building products at scale, leading engineering teams, and more recently, coaching people on how to master AI. I’m also the founder of Gud Prompt, a platform to save and share your AI prompts.
Let’s dive in for this week’s goodies. 💪
💡 The Spark
AI coding just hit a new gear. GitHub’s new Agents Panel lets you delegate tasks to Copilot from any page on GitHub, while Google’s Jules shipped a quality‑boosting critic update. AI that reviews its own code before it reaches you. Add Gemini 2.5 Pro inside Copilot, and the vibe‑coding future feels very “mind‑bending.”
🚀 The Build
This week’s headline move is Google’s Jules going all‑in on reliability.
Jules is an asynchronous coding agent that works in the background and opens pull requests you can review; no surprise edits to your main branch. It recently rolled out a critic that adversarially reviews AI changes to catch bugs and inefficiencies before you ever click “merge.” See the breakdown from ITPro and the GA context from InfoQ.
Code you can trust. The critic brings a second pair of (robot) eyes to every change, aligning with how teams already do code review.
Vibe‑coding at scale. Because Jules works via PRs, your flow stays intact while it drafts tests, bumps versions, or patches issues.
Plot twist: AI is now critiquing AI: a self‑review loop that levels up quality.
Zooming out: GitHub’s Agents Panel removes friction so delegating to agents feels like assigning a teammate. Also, Copilot’s Gemini 2.5 Pro option boosts reasoning for complex refactors.
Culture is catching up too: WIRED’s deep‑dive on “vibe coding” shows how companies like Notion pair humans with AI to ship features faster.
🛠️ Try This Wednesday (10‑minute build)
Project: Personal “Quick‑Habit” tracker you just built with AI assistance.
You’ll use: Replit
Start a Replit → Create a new Python project.
Prompt the app (paste this):
"Create a simple CLI habit tracker with commands: add, done, list, and stats. Persist data to a local JSON file. Include input validation and helpful error messages.Add a --today filter and a --week summary."Ask for tests:
"Generate unit tests for all commands. Include edge cases (empty input, duplicate habits, invalid flags)."Run an AI critique pass (Jules‑style):
"Critique this code for reliability and security. Suggest fixes and explain trade‑offs in plain English."Customize the vibe:
Add
remindcommand that prints a motivational message.
Swap the JSON store for a single
data/folder per user.Print a celebratory message when streaks hit 7 days.
Ship it:
Use Replit’s “Deploy” to package a simple CLI.
Share your repl link on X with #VibeBuilt and challenge 2 friends to beat your 7‑day streak.
✨ Vibe Check
TELUS rolled out Claude‑powered workflows to speed up internal innovation across teams, proof that large enterprises can vibe‑build faster with AI assistants.
Read the case study -> TELUS + Claude
ChatAndBuild is using Claude to democratize AI development globally, bringing agentic app‑building to non‑technical founders.
See how -> ChatAndBuild + Claude
From zero to demo in a weekend: one maker documented building an app in 48 hours with AI tools during a Lisbon hackathon.
Follow the journey -> Built in 48 hours
🔧 Tool Drop
🆕 Jules. Free‑tier, PR‑based coding agent now with a self‑reviewing critic to catch issues before merge.
🔥 GitHub Agents Panel. Delegate AI tasks from anywhere on GitHub without breaking flow.
⚡ Gemini 2.5 Pro in Copilot. Higher‑reasoning model option inside Copilot Chat for complex refactors.
🧠 JetBrains AI Assistant: Next Edit Suggestions. Smart “what’s next” prompts land in your IDE.
💫 Impossible Made Possible
PRs while you sleep: With Jules, you can offload version bumps, flaky test fixes, and boilerplate to an agent that proposes clean, reviewable PRs, no context‑switch.
Culture shift at Notion: WIRED embedded with a team adopting vibe‑coding to ship real features faster: human + AI pairing that feels like magic.
Listen in -> WIRED: How to become a vibe coder
🎯 Next Vibe
Next week, we’ll decode agentic programming patterns (plans -> actions -> critiques -> PRs) and share ready‑to‑paste prompts for production‑grade reliability.
Have a win to share? Post with #VibeBuilt and you might headline our next Vibe Check.
Keep building. The revolution favors the bold.
Found this helpful? Share with a fellow dreamer! I’m here to help turn your ideas into reality… let’s keep vibe coding!
So you like what you see? Then why don’t you…