😸 Devs on steroids

Your sunday scaries cure 🫶

gm legends and happy Sunday! I hope you're having a chill one before we all step back into the grind tomorrow. If you're still grinding, more power to ya! In today's weekly roundup: an AI bonanza in the Leaderboard highlights, breaking down China's big AI moves, and some news about developer's productivity.

The future of Intelligent Observability just dropped, and if you missed it live, don’t worry—we’ve got you covered.

Watch on-demand sessions featuring top industry experts as they break down AI-powered monitoring, system performance boosts, and all the major releases announced. Prefer a hands-on approach? Explore interactive workshops that walk you through New Relic’s latest innovations in real-world scenarios.

Get Caught Up

Weekly

Leaderboard highlights

	Twos PALs — AI-powered lists that save you time Twos Pals is a shared list app that makes it easy to jot things down and collaborate with friends, whether it's planning trips, swapping recs, or just remembering stuff together. It’s like Notes, but actually fun and social.	▲ 729
	Epiphany — Turn voice notes into instant actions in your favorite tools Epiphany is a minimalist journal designed to help you reflect with short, guided prompts and build a consistent writing habit. It keeps things simple so you can focus on what matters—your thoughts.	▲ 673
	Claude Web Search — Claude can now search the web Claude Web Search lets you ask questions and get AI answers backed by real-time web results. It’s like chatting with Claude, but now it actually knows what’s happening today.	▲ 610
	OpenAI GPT-4o Audio Models — Build Powerful Voice Agents OpenAI’s new GPT-4o audio models let you build apps that understand and talk—instantly. It's near real-time voice interaction, with emotions, interruptions, and all the messy magic of human conversation.	▲ 400
	Gemini Canvas — Interact and collaborate with AI to create Gemini Canvas is a collaborative space where you can think, plan, and build with AI in a more visual, interactive way. It’s like having a smart whiteboard that actually helps you get things done.	▲ 238
Show More Products

FROM THE FRONTIER

Is China winning the AI race?

Manus AI is China’s latest entry into the race for autonomous AI agents, and it’s got people talking. Unlike typical AI assistants that need constant prompting, Manus runs in the background, handling complex tasks—like analyzing stock trends or ranking job candidates—without waiting for instructions. It’s built on a multi-agent system that breaks down big problems into smaller tasks, then executes them asynchronously in the cloud. In theory, it’s a step toward AI that works like an actual assistant rather than a glorified autocomplete.

But does it live up to the hype? Early tests show promise, but Manus has already stumbled—fabricating data in financial reports and lifting text directly from existing websites. That raises big questions about trust and oversight, especially as China pushes to make Manus a major player in its AI ecosystem.

FROM THE BREAKPOINT

Developers on steroids

JPMorgan Chase is betting big on AI-powered coding assistants, claiming they’ve boosted developer efficiency by up to 20%. That means fewer hours spent on boilerplate code, debugging, and documentation—AI now handles the tedious stuff so engineers can focus on high-value projects like AI and data-driven systems. In theory, it’s a win-win: developers write less grunt code, and the bank moves faster. But if AI is handling the fundamentals, what does that mean for junior devs trying to learn the ropes?

For JPMorgan’s 63,000 tech employees, AI isn’t just another tool—it’s becoming part of the team. The bank has already identified 450 AI use cases, with plans to scale to 1,000 next year, turning software development into an increasingly AI-driven workflow. That’s great for productivity, but it also raises questions

TRENDING IN THE FORUMS

Is MCP just a hype train?

Most AI tools forget everything the second you close a tab. Model Context Protocol wants to change that—with shared memory and smarter agents that actually remember what you’re doing.

That’s what Ilia Pluzhnikov asked. Anyone here actually using it?

One dev tried the local server and got nowhere—Claude could only trigger prewritten commands, nothing felt useful. But others got weird with it. Someone built a Blender agent that models scenes from text. Another wired Claude into their local file system to edit and write code with context. One even bought a domain from inside a chat.

It’s clunky. It’s early. But the thread? Full of ideas that make you want to try anyway.

Latest Emails from producthunt