Compound Interest: AI Fluency & The New Work Deal

Open today’s feed and three truths collide: tech execs shout “AI will kill every job”. Jason Clauss argues AI is merely climbing atop a heap of decrepit, sluggish tech that had to be overthrown anyway. Meanwhile, PM Kurt Schrader is already shipping twice as fast by pairing his teams with GenAI. Contradictions? No—just Tuesday in 2025.

AI as a Career and Growth Lever

Generative AI is no longer the hack-day experiment you demo at all-hands; it’s the main engine for career and product lift. In his latest “Almost Timely” brief, Chris Penn breaks down how to pivot fast: feed your work corpus to a model, let it surface patterns, and then co-author the next iteration instead of fighting the code. I took that advice literally. 

Just a few months ago I built a personal GPT that would pave the way for my current role. It ranked new job leads against a rubric of mission fit, personality, and learning upside. I started by building a unified professional profile using every personality assessment I’ve ever taken in NotebookLM.  Later, I layered industry outlook data and company insights into a set of measures as to what would fit me best. One prompt spits out a 0-100 “hell-yes score” and three pointed questions for the hiring manager. 

The Governance Gap

The real AI risk isn’t sci-fi sentience—it’s unremarkable, ubiquitous code that outruns the rulebook. Last week 404 Media unearthed logs showing ICE piggy-backing on Flock Safety’s 5,000-strong license-plate camera network, running thousands of silent look-ups with no public oversight. Across the Atlantic, composer-turned-AI-activist Ed Newton-Rex forced the UK Cabinet Office to admit its internal assistant “Humphrey” runs on OpenAI, Google, and Anthropic models—the very firms Parliament is supposed to regulate. The pattern is clear: deployment first, governance someday. If you build or buy AI, treat guardrails as part of the product, not post-launch homework. Three minimum standards:

1. Model card & provenance. Ship a living datasheet that lists training sources, intended uses, and known failure modes; version it like code.
   
2. Quarterly red-team drill. Invite outsiders to jailbreak, data-poison, and privacy-probe your stack; publish fixes.
   
3. User exit ramp. Provide a single-click opt-out or masking toggle plus a purge API so downstream vendors can’t hoard ghost copies.

Do these now and you’ll dodge both regulatory whiplash and customer backlash—because when the auditors finally arrive, “we were moving fast” won’t pass muster. 

This governance gap reflects a broader pattern: institutions struggling to keep pace with individual empowerment.  It underscores a crucial point for every professional: navigating this landscape effectively means treating AI fluency like compound interest. Begin with one repeatable workflow, iterate daily, and reinvest the saved hours. Waiting for the tooling to stabilize is professional negligence in 2025. Skip now, and compound interest works against you, not for you.

The Workforce Social Contract Reboot

This AI-driven acceleration isn’t just reshaping our tools; it’s fundamentally rebooting the workforce social contract.  Across every thread this week, the same drumbeat: the workforce is rewriting the deal. Lenny Rachitsky’s 8,000-person survey finds tech pros still optimistic, yet burnout and distrust of corporate missions sit at record highs; nearly half expect to quit within a year unless autonomy improves. While managers polish retention decks, mid-career operators are opting out. Leah Tharin runs the math: a one-person B2B practice can clear $300 K-plus ARR with nothing but a laptop, Stripe, and LinkedIn reach—no org chart required. Brian Clark zooms the lens wider, charting millions who are exchanging single W-2s for portfolio careers and location-independent gigs because the old social contract no longer pencils out.

If you lead teams—or sell into them—refresh your offer sheet now. Salaries and swag won’t cut it. Frame roles as renewable “tours of duty” with clear missions, sunset clauses, and up-skill budgets. Price work by outcomes, not hours. Build staffing pools indexed on skills so people can pop in for a sprint, notch wins, and move on. Autonomy plus growth is the new currency; pay it, or watch talent vote with their feet.

Signal Boosts

• Jason Clauss – “AI Is Climbing to the Top on the Corpses of Decrepit Technologies”
  Clauss argues AI’s surge owes less to superintelligence than to decades of slow, broken software begging for replacement everywhere today. (jasonsaidwhat.substack.com)
  
• Christopher Penn – “How to Use Generative AI to Pivot Your Career”
  Penn shows practical prompts and workflows to co-create with models, turning personal data into tailored career pivots and opportunities fast. (Chris Penn – Marketing AI Speaker)
  
• Lenny Rachitsky – “How Tech Workers Really Feel About Work Right Now”
  Rachitsky’s 8,000-worker survey reveals burnout, autonomy demands, and readiness to quit unless roles deliver growth and meaning for top talent. (Lenny’s Newsletter)
  
• 404 Media – “ICE Taps into Nationwide AI-Enabled Camera Network, Data Shows”
  Leaked logs show ICE querying Flock’s nationwide ALPR network, proving surveillance AI outpaces regulation, accountability, and public consent already today. (404media.co)
  

Do This Friday

Friday challenge: fire up this Compound Interest Beats playlist, open your go-to LLM, paste your latest product brief, and ask, “How can this land harder?” Note the model’s trims, clarifications, or killer hook suggestions. It won’t reply, “Nobody reads these anyway”—promise. Email me the most surprising upgrade you uncover. Your insights could feature in next week’s newsletter.

Gratitudes

Watching former colleagues, longtime friends, and brand-new voices email me about a line that stuck has been equal parts humbling and energizing. Three weeks ago, six hundred subscribers felt impossible; today, leaders are circulating these posts in exec chat threads. Your curiosity makes the keyboard clack each morning. Thanks for reading, questioning, and shaping what comes next. With me daily.

Madam I’m Adam