Cliff Weitzman, Founder and CEO of Speechify, recently joined Harry Stebbings on 20VC for one of the most practical episodes the show has produced on building a consumer AI company. Weitzman covers growth, hiring, advertising, and the role of AI in running a company now used by over 60 million people. Here are the ten lessons most worth taking away.
What Is the Volume of Work Principle and Why Does It Drive Everything?
Weitzman applied to 26 colleges when most students applied to six. He wrote 48 drafts of his main essay. He now generates 1,300 AI ads per day. The throughline is the same: in any domain where the outcome is uncertain, the person who runs the most experiments wins. He quotes Alex Hormozi directly: volume times leverage equals output. Volume is not a personality trait. It is a deliberate competitive strategy.
Why Did Weitzman Meet 100 Consumer Subscription CEOs in Person?
At a pivotal point in Speechify's growth, Weitzman identified the top 100 consumer subscription companies by revenue and flew around the world to meet the people running them. He connected with founders from Instagram, Twitter, Plaid, Grammarly, Robinhood, and Honey, many of whom became early Speechify investors. His rule for any new domain is the same: read 100 books on it, find 100 experts, and talk to all of them. Information compounds fastest when you get it directly from practitioners.
Where Does Real Growth Knowledge Actually Live Inside a Company?
One of the most practically useful observations from the episode is that the best growth knowledge in any organization does not sit with the CMO. It sits two or three levels down, with the person actually buying the ads and reading the daily data. Weitzman made it a practice to find those people specifically, and brought that insight back to Speechify as a management principle: no matter how senior someone becomes, they are expected to stay hands-on. He describes the alternative as being a fat general sitting in the back rather than a warrior who can still take out their sword.
How Does Speechify Test 1,300 AI Ads Per Day?
Speechify built its own proprietary AI ad platform after Weitzman decided that using the same tools as everyone else would eliminate any competitive edge. The system generates roughly 1,300 AI ads per day, posts them automatically across Meta, TikTok, and YouTube, tracks performance in real time, and moves budget toward whatever is winning. It functions like an evolutionary bracket: ads that beat the existing top performers get more spend, and everything else gets cut. The lesson is that at scale, intuition about what will perform gets replaced by a system designed to find out empirically.
Why Will Token Spend Outpace Salaries?
Weitzman said on the episode that Speechify is approaching a point where it will spend more on AI tokens than on employee salaries. "If you don't spend a thousand credits a day, I am disappointed in you." The point is that AI spend is not overhead. It is leverage. The companies treating it as a cost to minimize are making the same mistake a pre-internet company would have made by limiting email use to save on server costs. Token spend that produces measurable output is one of the highest-return investments a fast-moving company can make right now.
What Is the Cutting and Bulking Framework?
Weitzman uses the bodybuilder's distinction between bulking and cutting to describe how companies should allocate strategic focus. You cannot grow revenue and cut costs at maximum efficiency simultaneously. Speechify ran profitable for four and a half years before entering its current hyper growth phase. His conclusion: any competent operator can cut costs, but growing revenue exponentially is genuinely hard, which is why investors reward it. Know which phase you are in and commit to it fully.
What Makes a Single Ad Generate Millions?
One ad generated approximately three million dollars in revenue for Speechify. Weitzman's process was to study the 100 best performing ads in history, rewrite each script to be about Speechify, and test dozens of variations. His insight about what makes an ad win: it is not the first three seconds. It is the first 1080 pixels, the single opening frame that either stops a scroll or does not. The winning ad was Weitzman in a suit and red headphones sitting in a hot tub. The ad he expected to win got 300,000 views. You cannot know in advance. You can only build a system that finds out faster.
How Should Founders Hire for Senior Roles?
Weitzman gave a newly hired head of growth seven days to demonstrate hands-on capability: making ads, editing creative, buying the ads themselves. The framework is borrowed from Keith Rabois: hire barrels, not ammunition. Ammunition needs direction. A barrel goes from zero to ten independently, full stack, without needing someone else to close the loop. Speechify looks for three things in every hire: genuine passion for the product, loyalty to the team, and the ability to learn fast and ship to production.
Why Does Only Conversion Matter?
Weitzman is direct on this: only do things that result in actual conversion. Speechify has a large TikTok presence with strong view counts and weak conversion back to the core product. His take is not that TikTok builds brand value. His take is that if it does not convert, it does not count. The ad spend is covered by people who pay. Everything else is secondary.
What Is the Real Lesson of the Dyslexia Story?
Speechify exists because Weitzman needed it. He could not read fast enough. He built a deep learning text-to-speech model in 2015 that now reads more than 10 million books worth of content to people every year. The clearest explanation for why the product works at 60 million users is that it was built to solve a real problem its founder had, seriously enough that it became a real product for everyone who shares that problem. That is still the most durable foundation a consumer company can have.
The full episode is available on YouTube. Speechify is available at speechify.com.
