Back to Podcast Digest
How I AI··51m

The Memelord built an API so agents can make memes for you | Jason Levin

TL;DR

  • Memelord is betting that agents—not humans—will become the power users of meme software — Jason Levin says “no UX is the best UX,” and built an API plus downloadable skills so tools like OpenClaw can generate brand-safe-ish memes from trending formats without a person fussing over whether they’re funny.

  • Levin got Memelord to $100K ARR on Bubble before hiring engineers — he started with a $6.90/month meme-alert newsletter that linked to Google Slides, then scaled a no-code app with 395 Bubble workflows purely through obsession before later moving the core product to Cursor-assisted engineering.

  • His strongest growth play is simple: replace lead-gen PDFs with weird little AI tools — Memelord’s free tools like a GigaChad maker, Steve Jobs portrait generator, and Snapchat caption builder have collected “hundreds of thousands of emails,” even going viral in Turkey from one TikTok about a filter.

  • Every marketer at Memelord is expected to vibe code — Levin’s argument is that handing creative ideas through layers of product and engineering is too lossy, so marketers should build the tiny “baby products” that drive demand directly on the marketing site.

  • The most memorable demos weren’t memes—they were hyper-personal personal tools — Levin built a Raspberry Pi bedside keyboard that routes half-asleep ideas to email or Linear via Zapier, and is now exploring AI systems that review his calendar, flag meetings that should’ve been emails, and turn real conversations into draft content.

  • On humor, his view is nuanced but practical: AI can be funny now, especially with the right models and prompts — he names Grok and Gemini as the funniest mainstream models, argues memes are the opposite of slop because they’re hyper-contextual, and says you need to push models into a more “unhinged” tone to get good comedic output.

The Breakdown

The pitch: meme software for an agentic internet

Jason Levin opens with the thesis behind Memelord: the internet is getting more entertaining, and brands that want attention need to get funnier, faster. His demo shows OpenClaw calling the Memelord API to generate memes off live topics like Iran, Claude, and “memetic warfare,” using a trending meme database instead of just spitting out generic slop.

Why Memelord may get an inflection point from agents

ClaVo points out the deeper story here: agents are oddly great meme users because they don’t get self-conscious or stuck on “cringe mountain.” Levin agrees and frames the company around a hard product truth he heard from Ramp’s CTO: “no UX is the best UX,” which makes his polished onboarding almost a bridge to the real future—the API.

From newsletter and Google Slides to a real product

Levin explains that Memelord started as a $6.90/month newsletter sending the newest memes, with links out to a Google Slides deck because he couldn’t code. That evolved into a Bubble-built app he scaled to $100K ARR without engineers, complete with 395 workflows and enough chaos that he got rate-limited on day two before he even knew what rate limiting meant.

Bubble scars, Cursor habits, and the rule that everyone builds

After raising money, he hired engineers, but he still codes through Cursor and requires every marketer at Memelord to vibe code too. His team uses that freedom to ship weird lead magnets fast—free tools, mini-generators, and experiments—because translating a marketer’s idea through engineering layers kills the spark.

The free-tools strategy that actually drives demand

One of the clearest takeaways is Memelord’s belief that “free tools are the new PDFs.” Levin says dumb little utilities like a bust-down filter, GigaChad maker, Steve Jobs portrait generator, and Snapchat caption tool have generated hundreds of thousands of emails, and one even blew up in Turkey after a kid posted it on TikTok.

Tools he’s bullish on: Linear, PostHog, and talking to models all day

Levin shows his commit graph turning dark green once AI-assisted coding clicked, then shouts out Linear and PostHog as standouts. He loves Linear’s craft and agent-friendly substrate, and says PostHog’s AI went from weak to shockingly useful—good enough to build dashboards from natural-language prompts like “show retention from podcast traffic.”

The best segment: a bedside keyboard for half-asleep ideas

Instead of bringing his phone to bed, Levin built a $10 keyboard-plus-Raspberry-Pi setup that lets him type ideas in the dark and send them to email or Linear through Zapier using prefixes like “lin-eng.” It’s hilariously overengineered and deeply relatable: he built it because he didn’t want to wake his wife, and because notebook ideas kept “getting lost in the sauce.”

Calendar agents, content mining, and the closing take on AI humor

He’s now using agents to review his week, suggest when he should skip engineering standups, and eventually flag meetings that should have been emails. On humor, he says AI absolutely can be funny now—especially Grok and Gemini—but only if you push it toward your voice; memes, in his words, are the opposite of slop because they’re dense with context, and that human-context loop still matters.