// Morning!
This week I’m digging into where AI actually works for operators, sharing field notes from my advisory calls, and dropping a few gems — from a paranoid CEO story to OpenAI’s own reflections, and even a one-click prompt upgrade.
For any new readers, welcome to Signal // Noise — the newsletter read by founders, CEOs, execs, and scrappy builders every Thursday. No buzzwords, no bullshit.
While I Was Writing Today’s Signal // Noise:
Want the whole vibe? The running playlist is right here. 🎧
From sound to signal, let’s get this baby rolling with what’s on my mind…
The Signal
One big idea, insight, or take - grounded in the real work, not theory.
AI Isn’t Eating Jobs (Yet), But It Should Reshape Margins: Where Operators Can Actually Win in 2025
AI hype is at peak volume, and to be honest, much of it is well-warranted.
Over the past three years, the cost of intelligence has dropped by over 10× every year, even as model capability has roughly doubled every seven months.
But for companies between $1–$20M in revenue, I don’t think the signal is about replacing people as much as it is about squeezing smarter margins. Operators who use AI to reduce CAC, improve support efficiency, and build a more defensible IP will see real ROI; peeps who just chase “AI everywhere” without a whole lotta discipline are honestly just gonna burn themselves out.
—> Example: in this McKinsey report, The State of AI: How Organizations Are Rewiring to Capture Value, it notes more than three-quarters of orgs say they use AI in at least one business function; yet over 80% report no material impact on EBITDA.
This disconnect between adoption & impact is exactly why chasing squirrels alone won’t cut it.
Where AI Actually Works Quickly
Some functions give you faster, more reliable wins: customer service, marketing, IT, finance.
These tend to have structured datasets (tickets, transaction logs, campaign metrics), making them easier to automate and measure. This Built to Scale study found firms that fully modernized with AI-led processes saw 2.5× revenue growth and 2.4× higher productivity compared to peers. (These stats underscore how choosing the right function matters.)
The truth is that AI can really help when it’s actually embedded in workflows and scaled, not when it’s just a side project or something your boss keeps yelling about at company all-hands.
Absolute musts: clean data, strong leadership, and clear KPIs.
The Playbook: Smart Moves vs. Pitfalls
Here’s where your leadership actually matters the most.
Empty mandates like “Just use AI more” or “We’re not hiring, so just figure it out with AI” show scattered thinking and a lack of vision.
To avoid ending up with hype over impact, try to lean into this structured checklist:
Smart Moves:
Tie to P&L: link every project to revenue or cost impact (pick a number, ex: drop support costs by 15%, cut CAC by 10%, etc.)
Start with clean data: choose a function with clean and structured data
Own it at the top: Founder/CEO/COO must advocate and track results
Plan to scale: design pilots knowing later you can scale for integration
Pitfalls:
Scatter experiments: spread thin, learn nothing
Issue vague orders: “use AI more” without KPIs is a waste
Ignore data hygiene: messy data kills means you don’t actually know what’s working
Outsource strategy to tools: AI needs human oversight and domain expertise
Why It Matters
Margin expansion is one of the strongest levers you have within your control.
As talent costs rise and competitive pressures intensify, the ability to extract 5-15% improvements in support, CAC, revops, etc, can be the difference between a plateau period & a growth stage.
Proprietary workflows, insights from your data, and speed that your competitors can’t match is as much about defensibility as it is about efficiency.
Field Notes
Dispatches from the field - lessons, stories, interviews, experiments.
Founders don’t post the questions that keep them up at night. But they do ask them on my calls.
Five quick Q&A takeaways from this week:
Sales vs. Product → Keep selling unless delivery truly breaks.
Founder Pay → Pay enough for sanity, not so much it starves growth.
Benefits → Even a roadmap signals maturity and trust.
Go-to-Market → Group wins? Celebrate, then chase repeatability.
Product Validation → “More” isn’t clarity—pilot before pricing.
A few Jawns to Check Out
Smart reads, sharp tools, or internet gems.
📰 Fresh POV | Nerd Sniping at Scale
A rare peek inside OpenAI’s own reflections on rocketing from 1k → 3k employees. What broke, what held, and why “nerd sniping” became a surprisingly useful lens on getting your project resourced.
🎧 Sweet Pod | “I am a complete f*cking animal”
Caryn Seidman-Becker on rebuilding CLEAR from bankruptcy into a multi-billion biz. She hits on TAM imagination, margins and FCF, optionality, staying “decently paranoid,” and even leading from the front by taking her salary to zero. I really enjoyed this interview — she is truly, a f*cking animal.
📕 Smart Hack | Gemini’s “Improve prompt” button
A 10-second trick that takes a meh prompt to a great one.
How: 1) Paste your prompt 2) Scroll down and click Improve prompt 3) Wait ~10s, copy the upgraded prompt, run it. Literally takes my average prompts & brings up to a 10/10.
Enjoy the weekend, and hit reply to tell me where you’re actually seeing AI move the needle the most.
Until next time, thanks for reading.
Jordan
P.S. Wanna work on something? Got a pod or content idea? → Email me | Need 30–60 min of advice? → Book here | Want a coach in your corner? → More info
