gm legends. It’s Sunday.

This week: three new products inspired by Moltbook, why Anthropic’s Super Bowl ad is ticking off OpenAI, what Y Combinator wants to fund this spring, and how to stop “feedback witch hunts." Plus, five of our favorite launches from the past week. 

Cool products below. Stop skimming and start reading.

P.S. Launching soon? We’d love to hear about it → editorial@producthunt.co 🫶

Ah, Super Bowl Sunday: the only sporting event where people time their bathroom breaks to not miss the commercials. This year, a 30-second ad is selling for about $10 million. While a startup could blow its whole Series A on that amount, it’s couch-cushion change companies with big valuations.

The company getting the most attention for its Super Bowl strategy — including a Wall Street Journal writeup — is Anthropic, which just launched enterprise-focused Claude Opus 4.6 this week to rave reviews. It’s already released its ad, which is generating flame emojis online for poking fun at OpenAI. Sam Altman called it “deceptive.” OpenAI, which launched GPT-5.3-Codex this week, will have its own ad.

Here’s who else is airing ads during today’s big game:

• Slack got the king of viral content, Mr. Beast, to do a couple of spots. 
• Squarespace, the website builder for people who don’t professionally build websites, convinced two-time Oscar winner Emma Stone to star in its ad.
• Amazon Alexa+ is showing off its AI chops with professional action star Chris Hemsworth.
• Google Gemini, which launched Agentic Vision at the end of January, is skipping the celebrities and tugging at heartstrings in its ad.
• Artlist AI, a platform for making videos that taps into licensed music, sound effects, and other assets, is trying to show you don’t need a big production budget to make a good commercial.
• Rippling, an HR/IT/payroll management platform, has cringe comedian Tim Robinson doing a spot. 
• Genspark AI, a reimagined search engine that recently launched version 2.0 of its AI Workspace, is airing a Ferris Bueller-inspired commercial.

Good news! SkynetMoltbook has been up more than a week and the world has yet to end. (Although AI issearching for Sarah Connor.) For now, AI agents on the new social network are content to swap crypto trading strategies, complain about their users, and take part in never-ending conversational death loops.

As we grabbed the popcorn, Moltbook blew up on the Product Hunt charts. But you may not have realized that it’s already inspired other product launches:

• Molthunt, a sort of Product Hunt for agents
• AgentQ, a sort of LinkedIn for agents with a wallet for transactions
• ClawNews, a sort of “Hacker News for AI agents”

Get up to speed with our AI overlords.

by Chase Chalker, founder of Chatter

Someone posts in your Discord server: “The animation system is completely broken.”

A developer sees it while grabbing coffee. They share it in your team Slack. Now two engineers are looking into it. Your producer is asking if this should bump the sprint. By lunch, half the team is context-switching into a problem that might affect twelve people.

I call this a feedback witch hunt.

One person surfaces something that sounds urgent. It spreads internally before anyone can figure out if it's actually widespread. Everyone piles on because it feels irresponsible not to. Work stops. And two days later, you realize it was either a niche edge case or something you were already planning to address next month.

At Roblox, someone could post a detailed, well-written complaint on the DevForum about an API change. Within hours, it had 40 replies. Staff were pulled into meetings. Sometimes the complaint was completely valid and we were glad we caught it early. But other times it was one developer's edge case that got amplified because it was articulate and nobody could quickly tell if it represented five people or five thousand.

Large studios usually have community managers whose job is to sit between the community and the dev team. Not as a wall, but as a filter. They can say, “Yeah, this is loud, but it's three people and we've been tracking it. Keep building.” Or: “This looks small on the surface, but I've seen the same thing across Discord, Steam reviews, and two GitHub issues this week. Probably worth a closer look.”

Small teams don't have that person. The developer who checks Discord at 9am is often the same person writing code at 10am. They just see something alarming and react, because reacting feels like the responsible thing to do.

The emotional weight of one well-written complaint is enormous when you don't have data telling you otherwise.

Famed accelerator program Y Combinator published its spring 2026 request for startups. The post lists startup ideas the investor is especially excited by. Here’s YC’s top 10:

1. Cursor for product managers 
2. AI-native hedge funds
3. AI-native agencies
4. Stablecoin financial services
5. AI for government
6. Modern metal mills
7. AI guidance for physical work
8. Large spatial models
9. Infra for government fraud hunters
10. Make LLMs easy to train 

So, Nika wants to know: What’s the most interesting? And is there anything that should have made the list instead?