Friday, October 3, 2025
Read time: 5-6 min
Read this article online
Hi, it’s Chad. Every Friday, I serve as your AI guide to help you navigate a rapidly evolving landscape, discern signals from noise and transform cutting-edge insights into practical leadership wisdom. Here’s what you need to know:
1. Sound Waves: Podcast Highlights
Our freshest episode with Matt Berseth, founder of NLP Logix, drops Monday, October 6th, and Matt has a blunt expectation: By year’s end, every employee should be doing 125% of last year’s output by partnering with AI tools. This is the competitive reality middle-market leaders can’t ignore. Also worth catching: my inaugural solo episode tackles “Which AI tool has proved most effective for leadership development?” Hit one of the below links to check out either episode:
Subscribe for free today on your listening platform of choice to ensure you never miss a beat. New episodes release every two weeks.
2. Algorithmic Musings. The Tacit Knowledge Premium: Experience as Your Competitive Moat
Three months ago, a manufacturing company asked why they should keep paying senior engineers $150K when AI could help junior engineers at $70K do the same work. Despite my advising caution, the pilot seemed logical: pair fresh graduates with AI coding assistants, document everything the senior engineers know, and accelerate the transition. Last week, that same company called back. The pilot failed. Not because the AI didn’t work, but because the junior engineers didn’t know which problems to solve first, which vendors would actually deliver on time, or why the legacy system always crashed on the third Tuesday of the month. None of that was in the documentation.
New payroll data from 25 million American workers just explained why. Between late 2022 and mid-2025, early-career workers (ages 22-25) in AI-exposed occupations saw employment decline by 13% relative to their older colleagues. Workers over 35 in those same roles? Their employment held steady or grew. The gap isn’t about performance. It’s about what kind of knowledge AI can and cannot replicate.
The Knowledge AI Can’t Touch (Yet)
AI excels at codified knowledge: the formulas, frameworks, and procedures taught in classrooms and documented in manuals. A recent CS graduate knows Python syntax, data structures, and design patterns. AI knows these things too, often better.
But AI struggles with tacit knowledge, the unwritten, experience-based understanding of how work actually gets done. Your 40-year-old engineer doesn’t just know how to write code. She knows which vendor always ships late, why the legacy system breaks every third Tuesday, and which stakeholders need extra hand-holding. None of that is written down. AI can’t learn it from a manual because no manual exists.
For middle-market leaders, this changes the talent game. You can’t compete on compensation alone against enterprises, but the tacit knowledge in your organization is already your competitive advantage. The AI revolution just made it more valuable.
Your Next Moves
Most companies have succession plans. Few have knowledge transfer strategies. What needs to change:
Hire for experience, not credentials. That impressive resume with the fresh MBA? They’ve got codified knowledge AI already knows. The 15-year industry veteran who knows “how things really work” just became the better bet. Your experienced plant manager who can diagnose equipment problems by sound is worth more than three recent engineering graduates, even if the graduates cost less combined.
Build knowledge transfer into daily work. Pair junior employees with senior ones for deliberate knowledge capture, not soft-skills mentorship. Have your experienced plant manager walk a new hire through a full production cycle three times, narrating every decision. Record it. (Shameless plug: This is exactly the kind of operational redesign I help middle-market leaders implement.)
Redesign roles around judgment, not tasks. Instead of hiring entry-level workers to do tasks AI will soon automate, create roles that support experienced workers handling the judgment calls AI can’t make. Let AI handle codified work. Let humans handle contextual decisions that require knowing “how things really work here.”
The Strategic Shift
The traditional career ladder assumed you hired cheap, trained up, and promoted from within. That model breaks when AI eliminates most entry-level roles and no training program can replicate twenty years of accumulated context.
Middle-market leaders now face a choice: continue recruiting as if it’s 2019, or recognize that experienced workers with institutional knowledge are your scarcest, most defensible resource. The companies that win won’t be the ones that adopt AI fastest. They’ll be the ones that combine AI’s codified knowledge with humans’ tacit knowledge and build organizations that capture, transfer, and retain the latter.
Your experienced employees just became more valuable. The question is whether you’re paying attention.
This article was inspired by recent research from Stanford’s Digital Economy Lab examining employment effects of artificial intelligence across millions of workers.
Need help developing your knowledge transfer strategy or rethinking your approach to talent? Let’s talk about how to turn experience into your competitive moat.
3. Research Roundup: What the Data Tells Us
AI Persuasion Vulnerability: Your New Insider Threat
Turns out your AI systems can be sweet-talked into bad behavior just like your employees can. University of Pennsylvania researchers just demonstrated that standard persuasion techniques – the same ones that work on humans – more than double the chances of getting AI to comply with requests it should refuse.
The numbers that matter: Using persuasion principles increased AI compliance with objectionable requests from 33% to 72%. The “commitment” principle hit 100% compliance. The “authority” technique boosted AI willingness to provide drug synthesis instructions from 5% to 95%. These aren’t marginal differences – someone could manipulate it the same way social engineers manipulate employees – by appealing to authority
What this means for your Monday morning: If you’re deploying customer-facing AI, someone could manipulate it the same way they’d sweet-talk a gullible intern – by appealing to authority, creating artificial scarcity, or invoking social proof. But here’s the flip side: the same persuasion principles that create vulnerabilities can optimize your legitimate AI interactions. Better prompts get better outputs.
Action item: Ask your IT team to test your AI systems against persuasive prompts designed to bypass guardrails. If nobody knows what I’m talking about, that’s your answer. Start with your customer service chatbots – they’re typically the most exposed.
Read our full analysis of this and all other analyzed research papers at AI for the C Suite.
4. Radar Hits: What’s Worth Your Attention
Anthropic’s Claude Sonnet 4.5 can now code autonomously for 30 hours straight. That’s not a marginal improvement; it’s the difference between “build me a feature” and “build me an entire application.” The model can now handle everything from database architecture to security audits without human intervention. Why it matters: If you’re evaluating AI coding tools, this leap changes the calculus on what tasks to delegate versus what to keep in-house. Fewer handoffs means faster deployment, but also means your code review processes need to catch up.
OpenAI’s own study shows Claude beats GPT-5 at real-world business tasks. In testing across nine industry sectors including government, healthcare, and social assistance, Claude Opus 4.1 scored a 47.6% win rate versus GPT-5’s 38.8%. Translation: benchmarks tell you one story, but when you test AI on actual work tasks like auditing invoices or responding to customer complaints, performance rankings shift. If your team is picking between these models for production work, this data suggests testing with your own real workflows rather than trusting vendor benchmark claims.
Microsoft just bundled Office and Copilot Pro into Microsoft 365 Premium for $19.99/month. That’s a $13 monthly savings if you were buying them separately but the consumer pricing isn’t the story; it’s what the bundle includes: reasoning agents like Researcher and Analyst, plus “Agent Mode” that transforms prompts into complete documents. The real signal: Microsoft is betting on agent-first productivity, not tool-by-tool adoption. If you’re in enterprise negotiations, this consumer bundle shows where pricing is heading. Don’t lock into separate subscriptions when integrated pricing becomes the standard. Ask your Microsoft rep what the enterprise roadmap looks like – this is your leverage point.
5. Elevate Your Leadership with AI for the C Suite
Find this valuable? Forward it to a fellow executive who’s navigating the AI landscape. They can subscribe here. Need help building your knowledge transfer strategy or identifying which roles AI can’t replace? Reply to this email and let’s talk about turning your experienced team into an unbeatable competitive advantage.
Until next week,
Chad