How I AIHow a non-technical founder built a $100K ARR meme company | Jason Levin (Memelord CEO)
CHAPTERS
Memelord’s thesis: entertainment wins, and memes are cultural transmission
Claire introduces Jason Levin and the Memelord premise: taking “being funny” seriously as a marketing advantage. Jason frames Memelord around the idea that the most entertaining brands win attention, and memes are the most information-dense unit of culture.
- •Why Memelord exists: make brands more entertaining to be more “likely” to win attention
- •Memes as cultural transmission and “memetic warfare” as a modern reality
- •Humor as a differentiator in a world trending toward generic AI slop
- •Early hint: Memelord evolves beyond a human UI into an API/agent product
Demo: Agentic meme creation via Memelord API + OpenClaw
Jason shows what it looks like when an agent generates memes on demand. The workflow pulls from a trending meme database, selects a relevant template, and writes captions—then iterates quickly with variations like “switch the caption.”
- •Agent prompts like “make a meme about X” produce context-matched templates (not random images)
- •Trending-meme database + caption generation as the core engine
- •Rapid iteration loops (e.g., caption swaps) as the real productivity unlock
- •Why this matters for marketing: narrow windows for trends, speed beats perfection
“No UX is the best UX”: building for an agent-first future
They discuss the shift from polished human interfaces to agent-driven usage, where the best experience is often just an API key. Jason shares how Memelord invested heavily in onboarding UX while knowing agents would eventually bypass it.
- •Ramp CTO quote: “No UX is the best UX” as a product strategy
- •Beautiful onboarding still matters for humans, but agents want direct interfaces
- •Investor anecdote: people don’t want to “use software,” they want outcomes
- •Packaging the API with agent “skills” becomes a new kind of product surface
From $6.90 newsletter to product: Google Slides MVP and meme alerts
Jason traces Memelord’s origin as a paid newsletter that sent meme alerts and linked to a Google Slides deck. The scrappy MVP validated the demand: people needed help staying current on memes and remixing trends fast.
- •Initial product: meme alerts + curated templates delivered as a newsletter
- •Low-tech delivery: Google Slides because Jason didn’t code
- •Core customer job: track trends + remix quickly for brand/context
- •Continuity of thesis: speed to trend is the competitive edge
Scaling without engineers: Bubble build to $100K ARR (395 workflows)
Jason explains how he grew to $100K ARR using Bubble, despite being non-technical. The tradeoff was complexity: hundreds of workflows that became hard to reason about—yet it proved the business before hiring engineers.
- •Built on Bubble pre-“vibe coding” era because waiting wasn’t an option
- •Reached $100K ARR without engineers; learned basics like rate limiting the hard way
- •Bubble’s power/limits: fast iteration but workflow sprawl and maintainability pain
- •Raising money later enabled hiring engineers while preserving speed culture
Cursor for non-technical builders + “everyone vibe codes” as policy
Jason describes how Cursor enabled him (and his marketers) to contribute directly, while engineers handle security-critical work. Claire highlights why Cursor is especially approachable: it teaches by letting non-technical people read and modify code safely with guidance modes.
- •Cursor workflows: Ask/Agent modes to explain, debug, and implement features
- •Division of labor: engineers on security/infra; founders/marketers ship front-end experiments
- •Company rule: every marketer must vibe code to reduce handoff lossiness
- •Value: faster idea-to-ship cycles and higher creative fidelity
Free tools as demand gen: meme utilities that captured hundreds of thousands of emails
Jason tours Memelord’s free tools section (lead magnets) and explains how small, weird utilities can outperform traditional PDFs. These tools went viral—especially in unexpected geographies—driving massive email capture and top-of-funnel growth.
- •Strategy: replace PDF downloads with interactive tools—often faster to build now
- •Examples: filters, generators, and mini utilities designed for shareability
- •Virality case: Turkey TikTok driving large-scale signups via a single tool
- •Operational mindset: minimal process; ship weird tools whenever inspiration hits
Let marketers cook—or they’ll leave: talent, autonomy, and the abundance mindset
They argue that AI makes traditional “prioritization” less relevant; organizations should bias toward shipping. Jason warns that restricting creative builders (especially marketers) leads to attrition—he left a prior company for that reason.
- •Handoffs destroy ideas; direct building preserves creative intent
- •AI shifts constraints: ship more, faster; “priority is yes” mentality
- •Token/compute spend as a strategic investment, not a line-item fear
- •Cultural takeaway: empower talented non-engineers or risk losing them
Tool stack and operating cadence: Claude/Gemini, Linear, PostHog (and agent-friendly workflows)
Jason shares the team’s core tools and why they matter in an agentic workflow. Linear and PostHog stand out because their AI layers and APIs let agents manage tasks and generate analytics insights without constant human UI usage.
- •Linear as an “agent task substrate”: opinionated workflows + useful AI
- •PostHog AI for dashboards and segmentation via natural language prompts
- •Multi-model reality: Claude/Chat/Gemini used where each is strongest
- •Pattern: humans stop logging into tools; agents operate them on their behalf
Build weird stuff IRL too: barbell strategy of digital agents + physical antics
Jason advocates pairing AI building with real-world creativity to stay grounded and generate differentiated content. He gives examples like Memelord CDs, domain hacks, and hosting quirky events to create memorable brand moments.
- •Barbell approach: extreme digital automation + extreme human/physical stunts
- •Physical artifacts as marketing: “API on a CD” as a story and hook
- •IRL event example: renting a theater to watch Instagram Reels
- •Broader point: human novelty becomes more valuable as AI content proliferates
Hyper-personal personal software: bedtime idea capture with Raspberry Pi + keyboard
Jason walks through a bespoke system to capture ideas at night without a phone, screen, or waking his wife. He built a keyboard+Raspberry Pi setup that sends Zapier-triggered actions (emails, Linear tickets) based on simple prefixes.
- •Problem: capture great ideas without doomscrolling or waking a partner
- •Solution: “keyboard with no screen” + Raspberry Pi + Zapier routing
- •Commands/prefixes to create Linear tickets or email notes automatically
- •Theme: build for yourself first; niche apps can be valuable even if tiny
Agentic calendar analysis: weekly review, time allocation, and meeting elimination
Jason shows how OpenClaw reviews his calendar weekly to summarize patterns and recommend changes. He extends the idea toward canceling meetings that could be emails and generating content ideas from real-life interactions logged in the calendar.
- •Weekly retrospective and week-ahead planning driven by calendar data
- •Insights: reduce attendance at standups; schedule more deep work
- •Future build: flag/cancel meetings that “could’ve been an email”
- •Content engine: turn meetings/DMs into draft posts while staying grounded in reality
Can AI be funny? Models, safety, and why “memes are not slop”
They debate AI humor: Jason believes top humans remain funniest, but AI is rapidly closing the gap with the right prompting and model choice. He distinguishes contextual memes from contextless slop and explains why many “safe” models underperform at edgier humor.
- •AI can be funny with the right setup, but humans still win at peak context + taste
- •Model ranking for humor (per Jason): Grok and Gemini ahead; Claude/ChatGPT feel safer
- •Memes vs slop: memes are high-context, information-dense communication
- •Practical approach: use AI to generate options; humans refine for taste and risk
What Jason doesn’t use AI for + closing advice: ship, cook, and stay human
Jason notes he avoids using AI to write in his own voice, preserving creative muscles like writing and stand-up. The episode ends with a call to embrace abundance, build fast, and maintain strong human craft alongside agent-first product surfaces.
- •Personal boundary: doesn’t use AI for his own writing/joke-writing craft
- •Prompting style: push models toward more unhinged vernacular when appropriate
- •Meta lesson: if you feel ahead and behind at once, you’re in the right place—keep building
- •Where to find Jason and Memelord; recap: entertainment + speed + autonomy wins
