Week 31 winners + a new way to win

Hey everyone,

Last week we changed how PostYourStartup picks its winners. Instead of counting upvotes, every launch now competes in the Startup Battle, and the first results are in.

🏆 Week 31 winners

Congrats to all three! As our #1, Psychostic gets a video made about it, which I'll post on Instagram and TikTok by the end of this week.

⚔️ How the Startup Battle works

• Head to head. You see two of this week's launches side by side and pick the better one. Every pick is a win for one startup and a loss for the other.
• One matchday a week. Each person gets one official matchday of 10 picks per week. You can keep playing for fun after that, but only your first 10 count.
• Best record wins. The startups with the best win percentage on Sunday at midnight Pacific win Gold, Silver and Bronze. A startup needs at least 20 games to qualify.
• What winners get: a Product of the Week badge for their site, a dofollow backlink, a spot in this newsletter, and #1 gets a video from me.

Last week, 52 people played 482 games. It was close, too: Silver and Bronze were separated by a single game.

Play this week's matchday →

💡 Don't build a product for developers in 2026

I just posted a video about this. Even before AI, I always said never build for developers, because they're the most likely to build it themselves instead of paying you. With AI, that's truer than ever.

A perfect example: Arcade Errors, which launched and competed right here on PostYourStartup, puts a playable retro arcade game inside your error pages. It's a fantastic product and I think it's great. I'd happily pay for something that cool, but as a developer my first thought was "I could build this in five minutes with Claude Code." So that's what I did. Now if you hit space on one of our error pages, you can play a startup-themed arcade game.

That's the problem with targeting technical people. They'll just rebuild your product and never pay. If anything, I'd argue Arcade Errors' pricing is a bit high for that reason: AI can replicate it in five minutes, and it did. So if you do target developers, price with the fact in mind that writing code is rapidly getting commoditized.

There are exceptions. Some developer platforms do something complex enough, with a real moat, that the average developer won't just vibe-code their own. And if a developer tool is the only idea you have right now, build it anyway, see how it goes, and learn from it. But for simple products, my advice is to target people who aren't technical at all. There are still tons of people who can barely use email or log into a web app, and they will gladly pay for products that make tech easy for them.

🤖 We built a ChatGPT plugin

I've been seeing reports on X and elsewhere of people getting serious growth from ChatGPT plugins, so we built one for PostYourStartup. It's submitted and in review with OpenAI right now.

Under the hood it's an MCP server (Model Context Protocol, the standard ChatGPT and Claude use to talk to outside tools). Version 1.0 is read-only: ChatGPT can search our startups, show what launched this week and the Startup Battle standings, find indie alternatives to popular products, list the best tools in a category, and suggest places to launch. No login needed. Next we'll add things like submitting your startup straight from ChatGPT.

If you're building one yourself, a few things tripped us up during submission:

• OpenAI's scanner is pickier than ChatGPT. It sent request headers our MCP library rejected, so the scan kept failing even though the plugin worked fine in ChatGPT. We had to accept looser headers.
• It speaks a newer MCP version than the official SDK. Its first request uses a protocol version the SDK doesn't support yet. That's expected: it falls back to the standard handshake, but it looks like an error in your logs.
• Label every tool explicitly as read-only and non-destructive (readOnlyHint, destructiveHint, openWorldHint), or the review flags it.
• Log what the scanner actually sends. The portal's error messages are vague; our server logs showed the real problem every time.

I'm also going to try this for my main product, Zinng, but we're going through the process with PostYourStartup first. I'll keep you posted on how it goes and whether we see a real jump in traffic from it.

Launching something? Post it on PostYourStartup and battle it out against this week's launches. It's free with our badge, or go live today with Premium. Submit your startup →

Tim