Welcome back!

This week we break down conversion data for free tools and trials, walk through Cursor's path to $100M+ ARR with zero marketing spend, explore a deployment-ready reasoning model, and show how ElevenLabs turned voice actors into a compounding sales force.

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

Founder’s Intel

Are free tools/trials part of your growth engine?

Data Intel

• Free tools: Can generate conversion rates up to 800% higher than standard signup forms.

• Freemium: Converts 2–5% of users to paid on average. Website visitors sign up for freemium accounts at 13.3%.

• Free trials: Opt-out trials (requiring credit card upfront) convert at 48.8%, while opt-in trials (no credit card) convert at 18.2%.

Why It Matters

Free tools and trials aren't giveaways, they're distribution assets. When a user gains real value with zero friction, they experience your brand’s promise firsthand, share links, and return with intent. Utility creates proof, shareability multiplies reach, and you get qualified top-of-funnel traffic at lower cost than paid campaigns.

Quick Tip

Extract one high-value feature from your product or build something that serves your audience and aligns with your brand, ship it as a free tool, make sharing frictionless, then track where users come from and how they engage. When it converts consistently, rinse and repeat.

Behind the Tool

From Side Project to Startup Vision Engine: How VenturusAI Found Its Path

The Spark

Alin Ciciu, a lifelong entrepreneur and self-taught developer, built VenturusAI to help founders clarify their vision and understand their businesses faster. While studying business and working full-time as a software engineer, he realized how inefficient traditional business analysis was. Using GPT as the foundation, he built an AI system that cut through the noise to help entrepreneurs define their path forward.

The Build

VenturusAI started as a side project, but Alin quickly saw its potential. The hardest decision wasn’t technical, it was knowing when to quit his job and go all-in. Early on, he resisted paid ads, opting for feedback-driven growth through Reddit communities and early users. “Your vision is never complete without feedback,” he says. “You need people who don’t care about you to tell you the truth.”

The Breakthrough

Once listed on TAAFT, user growth accelerated. VenturusAI gained steady organic traction and became a go-to tool for early-stage founders. Alin attributes this to its practical edge: building custom analysis tools on top of LLMs, not around them. “Anyone can wrap GPT,” he explains. “The difference is what you build on top; like tools that turn raw text into real business insight.”

The Next Chapter

Now two years in, Alin is expanding VenturusAI’s feature set with deeper integrations and intelligent agents capable of analyzing live business data. He believes the next five years will bring human-level AI agents that handle repetitive digital work, freeing people to focus on creativity and strategy.

Key Lesson

Build for usefulness, not hype. Products that solve real problems and prioritize user benefit over profit always outlast the noise.

Tool of the Week

Cursor's Zero-Marketing Blueprint

What’s Cursor?

An AI code editor where developers describe what they want in plain English, and AI writes, edits, and debugs code. Hit $100M ARR in ~12 months with $0 marketing spend.

What Worked

• Built audience before launch: Co-founder Aman Sanger posted AI coding content on Twitter for months, testing ideas publicly.

• Shipped two killer features after a flat period: Command+K (instant edits) and Tab (code prediction) took DAU to 30,000 by Jan 2024.

• Dogfooded ruthlessly: Entire team used Cursor daily as their main editor. Built what broke their own workflow, ignored everything else.

• Free tier hooked users: 2,000 AI completions per month let developers build real projects before paying. Not a demo, actual value that made upgrading inevitable.

Founder Quote

“The really magical thing about this product was that we were users of it. So we could iterate incredibly quickly, and we tried all these different things that summer. Then we found this core set of features that worked incredibly well.” — Aman Sanger, Cursor co-founder

Key Lesson

Build your audience while you build your product, test ideas publicly, then iterate based on how people react.

Fresh Out of the Lab

K2 Think

What Is It?

A lean open-source reasoning system (32B parameters) from MBZUAI and G42 that delivers world-class mathematical and logical problem-solving at a fraction of typical model sizes. Ranks among the top performers on competition-level benchmarks including AIME, HMMT, and Omni-Math-Hard.

What’s New

Built on six pillars of innovation: long chain-of-thought supervision for deep logical tracing, reinforcement learning with verifiable rewards for accuracy on hard problems, agentic planning to decompose complex tasks, and test-time scaling for adaptive reasoning. Fully open-source from training data to deployment code.

Why It Matters

Founders can build reasoning-first apps without massive infrastructure overhead. This unlocks smarter research assistants, math-heavy automation, competitive problem-solving agents, and complex decision systems that stay on-premise and under complete control. Reasoning performance that used to require models 20X larger now fits in deployable, transparent systems.

Founder’s Edge

This Week’s Builder Toolkit

• Dev Tool: Build and deploy full-stack web apps from a single prompt with Bolt.new to ship working prototypes faster.

• Free Dataset: Compare your growth rates, churn, and CAC payback against real SaaS benchmarks from OpenView.

• No-Code App: Turn your Airtable or Sheets into client portals and internal tools using Softr without writing a single line of code.

• Productivity Hack: Let RescueTime run silently in the background to find out exactly how much time you waste vs. build each week.

• Learning Resource: Learn AI strategy through AI For Everyone to make smart decisions about AI without becoming an engineer.

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

How ElevenLabs Turned Creators Into a Sales Team

Early on, ElevenLabs needed more high-quality voices to compete, but getting creators to trust a new AI platform was tough.

Instead of spending on ads or partnerships, they launched Voice Actor Payouts in February 2024 (a way users could upload a verified clone and earn money every time someone used it).

The results compounded fast: $1M paid out by October 2024, $2M by January 2025, and $5M shortly after. Over 5,000 voices are now live in the library.

Here's what made it work:

Every voice became a landing page that creators promoted to their own audiences. The economic incentive aligned growth on both sides. More creators joined because they could earn. More users signed up because the library kept expanding. ElevenLabs paid the people creating their scarcest asset, and those people brought the customers.

Key Takeaway

Find the scarcest thing users create in your product, then pay them when others use it. Turn your supply side into your distribution engine.

— Mati Staniszewski, ElevenLabs co-founder

Weekly Challenge

One Experiment. One Week. One Win.

The Goal

Get 10 pieces of user feedback that reveal one pain point you didn't know existed.

How It Works

1. Pick ONE feedback method (in-app survey, reply-to email, or 5-minute user calls).

2. Ask ONE open-ended question about their biggest frustration or unmet need.

3. Share your chosen feedback method and results in the TAAFT community by the end of next week.

Why It Works

Your roadmap should be built on real problems, not assumptions. Some of the best product ideas come directly from users.

Spotlight

We’ll showcase the most eye-opening discovery in an upcoming newsletter.

AI Market Watch

Deals, Discoveries, and Demand

Megadeals

• Liberate – $50M Series B (insurance AI agents)

• Dexory – $165M round (AI warehouse robotics)

• Lila Sciences – $115M Series A extension (AI science factories)

• Starship Technologies – $50M Series C (autonomous delivery robots)

New Research

• R-WoM (computer-use agents)

• Vgent (graph-RAG for long video)

• LaSeR (last-token self-rewarding RL)

• RIPRAG (attack black-box RAG with RL)

• Video Models, Zero-Shot Reasoning (video-first reasoning)

• LLM-guided Fuzzing for Agent Browsers (prompt-injection defense)

Search Trends

Top “best AI for…” searches:

1. Content detection

2. Text-to-video

3. Writing

4. Math

5. Writing essays

Thanks for reading!

If you’re building, experimenting, or scaling something new, we’d love to hear from you. Hit reply and share it with us.

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

See you soon,
— AI Empires