From Zero to 300,000 Users

How a weekend side project grew into a top AI tool

Hey there! It’s Thursday, and in this week’s issue, we’ve got onboarding tips that keep users hooked, a weekend project that turned into 300,000 signups, and an AI tool that went viral for being a little too bold.

Have suggestions? Send them to us at [email protected].

Letter from TAAFT Founder

👋 Hi, welcome to the first edition of AI Empires. I’m Andrei, the founder of TAAFT.

Over the past couple of years, we’ve worked with thousands of AI founders and helped them launch and promote their business. We want to share some of the learnings with you to hopefully help you avoid some common mistakes and even learn a few tricks.

We’re also going to be sharing some lessons we learned building TAAFT into the #1 website for AI tools. I hope you’ll enjoy this and feel free to reply to this email with your feedback or ideas for what we should be covering next.

Keep shipping! 🚀

Founder’s Intel

Is data-driven onboarding your growth multiplier yet?

Data Intel
  • 76% of SaaS companies that personalize onboarding see higher activation.

  • 63% reduce churn in the first 30 days.

  • 2x more users reach their “aha” moment faster.

Why It Matters

Onboarding isn’t a welcome tour, it’s where retention is won or lost. AI founders who build behavior-driven flows (think in-app guidance and triggered nudges) improve UX and convert trial users into loyal customers. Ignore this, and users will churn before they ever see your product’s true value.

At TAAFT, for example, one thing we do during onboarding is send users their first AI tool alert. That way they immediately see the value of personalized tool updates, which leads to higher open rates and removes confusion.

Quick Tip

Start small: track one activation metric, automate a single key nudge (like an email or in-app prompt), then scale from there. The compounding effect is huge.

Behind the Tool

From Zero to 300,000 Users: How ImageToCaption Took Off on TAAFT

The Spark

Joshua Heller, a former photographer, was tired of spending hours writing captions. One weekend, he asked ChatGPT for project ideas and landed on a solution to a simple problem: turning images into authentic captions. With no prior experience in cloud deployment or scaling, he built the MVP himself in two days and listed it on TAAFT.

The Build

The first version was rough, but demand was immediate. Traffic spiked and he was losing $100 a day on server costs. Forced to learn fast, he rebuilt from scratch, layered in a freemium model, and kept the product laser-focused on ease of use and brand-authentic captions.

The Breakthrough

Within weeks, ImageToCaption.ai crossed 100,000 visitors. Today, it has over 300,000 users and remains one of the highest-traffic tools on TAAFT. The edge? Captions that don’t sound like AI. As Heller puts it, “Generic tools sound robotic. Ours feels human and on-brand.”

The Next Chapter

Heller is expanding ImageToCaption into a holistic social media content platform: image generation, scheduling, and integrations, while also launching an AI consulting agency to help businesses build their own tools.

Key Lesson

Don’t overthink your first version. Launch fast, validate with real users, and improve from there. The traffic, feedback, and data will tell you what matters.

Tool of the Week

Cluely’s Viral Playbook

What’s Cluely?

An AI overlay that sees your screen, listens to your calls, and feeds you real-time answers without anyone knowing.

What Worked
  • Launched with a bold, controversial video (“Interview coder for everything”). The MVP barely worked, but it went viral anyway.

  • Built a content machine: ~50 interns posting hundreds of short-form videos daily. Rule: "Post 100 ideas, find 1 viral hit," then duplicate across every social channel.

  • Followed the data: 1B+ views guided product direction directly from user demand, with no guesswork involved.

  • Turned controversy into marketing: the “cheat on everything” pitch fueled debates and free publicity.

Founder Quote

“If your distribution is strong enough, you don't need to guess what users want. The data tells you exactly what to build.” — Roy Lee, Cluely co-founder

Key Lesson

Go big on viral content to validate your AI idea. Let the audience guide your roadmap and amplify your reach.

Fresh Out of the Lab

GLM-4.5V

What Is It?

A next-gen open-source vision-language model (106B-parameter MoE) that handles images, video, documents, and GUIs with SOTA accuracy across 40+ benchmarks.

What’s New

A “Thinking Mode” toggle that switches between fast responses and deeper chain-of-thought reasoning. Strong at chart parsing, GUI understanding, and visual grounding.

Why It Matters

Founders can now build agents that don’t just see but also reason. This unlocks smarter assistants, code-from-screenshot tools, automated report processing, and visual workflows without relying on third-party APIs.

Founder’s Edge

This Week’s Builder Toolkit

  • Dev Tool: Supabase is like Firebase, but open-source and way friendlier to your budget. It’s great if you’re racing to build your MVP.

  • Free Dataset: Get inspired by 500 Startup Pitch Decks and borrow a storytelling trick or two for your next pitch.

  • No-Code App: Automate repetitive tasks across thousands of popular apps with Zapier x ChatGPT so that you can focus on what really matters.

  • Productivity Hack: Use Hootz to block distractions, prioritize tasks, and keep you motivated with mood-improving wins.

  • Learning Resource: Practical Deep Learning is a free, hands-on course that’ll take you from zero to deploying your first AI model fast.

Note: If you’ve found a tool that’s sped up your build process, hit reply and share it. We’ll feature the best submissions in a future issue.

AI Founder’s Journal

The Free Demo That Put Supademo on the Map

In the early days of Supademo, growth was flat. Cold outreach wasn’t working, and it felt like nobody cared about the product.

Instead of giving up, Joseph tried a new approach. He went into startup communities and offered to create free interactive demos for anyone who asked.

On Reddit’s SaaS forum alone, his post racked up thousands of views in 24 hours. Founders started taking him up on the offer, real sign-ups followed, and many of those users converted into paying customers.

Key Takeaway

Lead with value where your audience already hangs out. Don’t pitch, let people experience your product for free and they’ll sell themselves on it.

— Joseph Lee, Supademo co-founder

Weekly Challenge

One Experiment. One Week. One Win.

The Goal

Increase an existing metric for your user base by at least 5% in 7 days.

How It Works
  1. Pick ONE high-leverage area (activation, retention, or expansion).

  2. Run ONE experiment (personalized onboarding email, AI-driven in-app tip, or VIP upsell offer).

  3. Share your experiment setup and results in the TAAFT community by the end of the week.

Why It Works

At scale, even small optimizations create huge compounding wins.

Spotlight

We’ll showcase the best growth experiment in an upcoming newsletter.

AI Market Watch

Deals, Discoveries, and Demand

Megadeals
New Research
Search Trends

Top “best AI for…” searches:

  1. Image generation

  2. Coding

  3. Writing

  4. Math

  5. Writing essays

Thanks for reading!

We hope you enjoyed this first issue and would love to hear your feedback. Hit reply and let us know what you think.

We’ll be back next week with more founder stories, fresh tools, and data worth your time.

See you soon,
— AI Empires