Cliff Weitzman: What I Learned from 100 of the World’s Top CEOs & Why Tokens Will Outspend Salaries

Advertising Strategy & Spend:

• Speechify tests nearly 1,000 AI-generated ads daily alongside 8,000 human-generated organic creatives.
• The company operates on a "meta-first" rule: do not spend on any platform other than Meta until reaching $100,000 in monthly ad spend on Meta.
• Speechify is one of 200 companies provisioned to test ads on OpenAI, prioritizing rapid learning over immediate profitability during the testing phase.
• Token costs for AI operations are projected to exceed actual salary costs for Speechify by next year.
• One specific YouTube ad generated $3 million in revenue; the team attributes success to repurposing viral formats (e.g., "American founder vs. French founder") and high-volume testing of 48+ video variations.

Product Development & AI Architecture:

• Speechify has built its own proprietary AI video generation platform to create ad variations (reskinning) without relying on external tools that everyone else uses.
• The company uses "Manus" and custom LLM agents to generate deep research documents, mapping medical anomalies for family members and analyzing business competitors in real-time.
• Jarvis, the company's voice agent ecosystem, is moving from text-to-speech to full voice agents for productivity and enterprise use.
• Speechify has reduced inference costs to single-digit dollars per million characters, a 10x to 100x reduction compared to competitors like ElevenLabs, enabling long-term profitability.
• The engineering team uses Claude Code (Anthropic) internally as their primary coding environment, enforcing a rule where engineers must utilize AI to ship features.

Hiring & Organizational Culture:

• Speechify prioritizes "Adversity Quotient" (AQ) over IQ or EQ, looking for candidates who can persist through 8+ hours of difficult problem-solving.
• Hiring criteria include high loyalty, the ability to learn fast, and the capacity to ship to production immediately.
• The company bans traditional performance reviews, replacing them with a "ship or fail" metric where credit is only given for code in production.
• Speechify enforces a "3-hour response rule" (the "Titanic rule") where all internal messages must be responded to within three hours to maintain speed.
• The company hires "barrels" (full-stack creators who can code, edit, and run ads) rather than "ammunition" (specialized workers who need management).
• Meetings are minimized to avoid bottlenecks; asynchronous communication via calls or Loom videos is preferred for decision-making.

Leadership Philosophy & Personal History:

• Cliff Weitzman applied to 26 colleges (the maximum allowed via a workaround) by writing 48 drafts of a single essay and attending academic workshop daily due to dyslexia.
• Weitzman's career strategy is built on "Volume x Leverage = Output," demonstrated by sending hundreds of cold emails to founders of top subscription companies until he secured early investors.
• The founder believes in "Bulking and Cutting" cycles: companies must focus on hyper-growth for several years before shifting focus to cost-cutting and profitability.
• Speechify has been profitable for four and a half years while growing from $5M to $10M+ monthly revenue without public fundraising disclosure.
• Weitzman personally solved dyslexia and ADHD by building the core AI technology, noting that 40% of billionaires share these neurodivergent traits due to developed resilience.

Investment & Market Views:

• Weitzman predicts NVIDIA will become a $10 trillion company, citing a P/E multiple of roughly 35x revenue as reasonable given growth.
• He is bullish on Solar as the immediate solution to AI energy demands, viewing fusion as a long-term play (2030s+).
• Weitzman favors Meta (Facebook) over TikTok for long-term equity, citing Mark Zuckerberg's acquisition prowess, data moat, and lack of direct LLM competition.
• Contrarian view: He would buy OpenAI stock over Anthropic despite using Anthropic products, citing OpenAI's superior operator (Sam Altman) and recruiting capability.
• He advises investors to "diversify to concentrate," identifying 1-2 high-conviction opportunities (like NVIDIA or Tesla in 2015) to bet on heavily.

Future Trends & AI Evolution:

• Weitzman identifies Quality Assurance (QA) as the most valuable skill in an AI-native world, as AI-generated code requires rigorous human testing to ensure reliability.
• He predicts the next 10 years will define a new human evolution where agency comes from the ability to process information faster via AI tools rather than physiological changes.
• He warns against wealth inequality exacerbated by AI, suggesting a government-provided basic smartphone for every citizen as a human right for economic participation.
• Weitzman advocates for a "teacher agent" model where AI tutors replace traditional K-12 schooling, focusing on competency-based assessment rather than seat time.
• He anticipates a "voice-first" era where text-based communication is replaced by voice agents for speed and efficiency in professional settings.