Kimi's Shock, Corporate Games
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Sunday, July 19, 2026
A $2 billion company's AI strategy was written by an executive who'd never once opened ChatGPT. Meanwhile, China's Kimi K3 rattled Wall Street enough to dip the Nasdaq, and a consultant's anonymous source confessed to rewriting entire codebases in a different programming language just to game an internal AI leaderboard. I've got the inside view on what's real, what's theater, and what you should actually care about this Sunday.
Today in AI:
- The Executive Who Never Logged In - A consultant revealed that a C-suite leader at a $2B+ company built an entire AI-centered technical strategy without ever having used ChatGPT or any AI tool. The anecdote anchors a larger essay arguing AI mania is warping corporate decision-making from the top down. Ludic Mataroa Blog
- Kimi K3 Spooks Wall Street - Chinese startup Moonshot AI released Kimi K3, an open-source model that independent benchmarks from Arena.ai and Vals AI say competes with flagship proprietary models. The Nasdaq dropped about 1% on Friday as investors sold off chip stocks like Nvidia. TechCrunch
- David Sacks Uses Kimi to Bash Anthropic - The former Trump AI czar seized on Kimi K3's release to argue the U.S. is losing the AI race to regulation and what he called Anthropic's "woke lobotomized models." Translation: a Chinese open-source model just became ammunition in an American political food fight. TechCrunch
- The Token Leaderboard Arms Race - An engineer at an unnamed company admitted to cloning an entire Go codebase and having AI rewrite it in Zig just to climb an internal AI usage leaderboard and keep their job. The metric is the mission now, and the mission is absurd. Ludic Mataroa Blog
- Qwen 3.8 Max Preview Goes Cheap - Alibaba's Qwen Cloud rolled out upgraded individual and team token plans with lower prices for its latest Qwen3.8-Max-Preview model, offering unified access to text, vision, speech, and image generation. The pricing war for API access just got another entrant. Qwen Cloud
- SQLite Gets an AI Explainer - Simon Willison built an interactive SQLite query plan explainer using Claude's Fable model, running SQLite via Python in WebAssembly directly in your browser. It won't replace knowing SQL, but it makes opaque query plans readable for anyone willing to paste one in. Simon Willison
- China's Open Source Play Gets the Subsidy Theory - Security commentator Daniel Miessler argued that China's cheap open-source AI releases are a deliberate CCP strategy to undercut U.S. lab valuations, weaken the stock market, and shift global power dynamics ahead of potential Taiwan moves. The theory is provocative and contested. Danielmiessler
Today's Takeaway:
Moonshot AI's Kimi K3 hit frontier-level benchmarks while Alibaba's Qwen slashed API pricing the same week, and the American response wasn't a technical counter - it was a political argument. David Sacks blamed regulation; Daniel Miessler blamed a CCP subsidy plot; an anonymous engineer quietly rewrote a codebase in a language nobody asked for just to look busy. The pattern isn't that China is winning or losing. It's that cheap, capable models from any source expose a specific vulnerability: companies that adopted AI as a strategy slide rather than an operational tool have no way to tell whether their spending is working. When an executive who's never touched ChatGPT writes a $2B AI roadmap (Ludic Mataroa Blog), and an engineer games a token leaderboard to survive, the threat isn't a foreign model - it's an organization that can't distinguish performance from activity. The firms that will weather a pricing crash are the ones whose AI use ties to measurable output, not internal theater. If your company's AI KPI is token consumption, you're measuring the wrong thing.
"The real AI threat isn't a foreign model - it's mistaking activity for output."
📋 Try This
From tech-bro culture in San Francisco to ethical consumption debates, this week's stories all share a common thread: people respond to things that feel real and relatable, not polished and distant. Use these prompts to make whatever you're writing or sharing actually connect with the people reading it.
For Business Owners:
I run a [TYPE OF BUSINESS] and I'm trying to connect better with [TARGET AUDIENCE]. Here's a piece of content I recently wrote or a message I recently sent: [PASTE YOUR TEXT]. Rewrite it so it sounds like a real person talking, not a brand. Remove any corporate-sounding phrases. Add a short, honest moment - something that shows we understand a real problem our audience faces. Keep it under [WORD COUNT] words and make the opening line something someone would actually stop scrolling for.
For Personal Use:
I need to write a [TYPE OF MESSAGE - e.g. email, social post, thank-you note, speech] for [SITUATION OR AUDIENCE]. Here's what I'm trying to say: [PASTE YOUR DRAFT OR DESCRIBE YOUR MAIN POINT]. Make it sound warm and human, like I'm talking to a friend. Remove anything that sounds stiff or formal. Add one small specific detail or honest feeling that makes it feel personal. Keep the tone [CASUAL / WARM / SINCERE] and the length short enough to read in under a minute.
💡 Copy either prompt, swap the brackets with your own details, and paste it into ChatGPT or any AI chat tool.
The Bottom Line
The Pattern: AI competitiveness is shifting from a model quality question to an organizational honesty question - whether companies can admit what's actually working versus what just looks impressive on a dashboard or an earnings call.
The Other Read: Kimi K3's Nasdaq dip was barely 1%, and Moonshot itself acknowledged it trails Claude Fable 5 and GPT 5.6 Sol. This could easily be another DeepSeek moment: a brief panic followed by business as usual once benchmarks settle. That said, the corporate dysfunction the release exposed is real even if the competitive threat is overstated.
Your Move: Pull up the SQLite Query Explainer Simon Willison built (link) this Sunday and paste in one query from your own database - takes five minutes. It's a small, concrete way to use AI as an actual learning tool instead of a token-burning exercise.
What We're Working On
✨ Founding Cohort Special - 60% Off! - Use code MAF20 to join for just $20/month (regularly $50). Get weekly group sessions & workshops, self-paced courses for all levels, access to tools & templates, challenges with peer feedback, and 24/7 support community. → Join Now
✨ Free 30-Minute AI Consultation - Discover how My AI Fluency can help your business unlock the potential of AI. We'll discuss your goals, explore practical AI opportunities for your industry, and outline clear next steps. → Schedule Free Call
✨ How AI-Fluent Are You? - Test your AI fluency with our interactive quiz. See how you stack up and discover what to learn next. → Take the Quiz
💬 Community | 📞 Book a Consultation | 🌐 Website
Fluently yours, The My AI Fluency Team