## Sources 1. [TBM 416: Investment Stewardship (As Habit)](https://cutlefish.substack.com/p/tbm-416-investment-stewardship-as) 2. [AI Is Here, But The Hard Parts Haven't Changed](https://joereis.substack.com/p/ai-is-here-but-the-hard-parts-havent) 3. [The data reliability question you're avoiding](https://andrewrjones.substack.com/p/the-data-reliability-question-youre) 4. [How I Built SEO Optimized Content Machine Using Claude Cowork and Apify](https://aimaker.substack.com/p/claude-cowork-apify-seo-content-machine) 5. [The Convenience Trap](https://jessicatalisman.substack.com/p/the-convenience-trap) 6. [TBM 412: Institutionalized Overload (Now With AI)](https://cutlefish.substack.com/p/tbm-412-institutionalized-overload) --- ### AI Is Here, But The Hard Parts Haven't Changed by Joe Reis * **Universal AI Adoption:** A March 2026 Practical Data Pulse Survey revealed that almost all data professionals use AI tools, with a broader survey showing 82% of respondents use them daily [1, 2]. Claude is the clear favorite, utilized by 49% of the community [3, 4]. * **The Speed vs. Quality Dilemma:** While 57% of professionals report that AI helps them write code significantly faster, this rapid generation raises concerns about a drop in quality and the creation of a new form of technical debt: code that works but that nobody fully understands [3, 5, 6]. * **Focus on Fundamentals:** Data modeling and semantic layers were identified by 49% of practitioners as the most critical focus areas for 2027 [4]. AI surfaces the need to get foundational elements right, highlighting that capturing organizational context and meaning remains notoriously difficult and cannot be automated away [7, 8]. * **The Unchanged "Hard Parts":** AI cannot solve core organizational bottlenecks like legacy systems, technical debt, poor requirements, and lack of leadership [9, 10]. Success requires investing in fundamentals, as technology alone cannot fix human-centric organizational issues and poor incentive structures [10, 11]. ### How I Built SEO Optimized Content Machine Using Claude Cowork and Apify by Wyndo and Gencay * **Automated Content Pipeline:** The authors detail a system that automatically researches and writes SEO-optimized articles using a combination of Apify and Claude Cowork [12, 13]. * **Data Scraping with Apify:** The workflow uses Apify's Google Search Results Scraper to analyze existing search results to find content gaps, and the Google Keyword Data Scraper to find search volume and ranking difficulty [14-17]. * **AI Execution with Claude Cowork:** Unlike a standard chatbot, Claude Cowork is a desktop application capable of executing tasks, reading and writing files, and running a specialized "SEO Writer skill" to draft articles based on the scraped JSON data [18-21]. * **Scheduling and Adaptability:** The system can be scheduled to run overnight so articles are ready by morning, and it can be triggered remotely via a mobile app called Dispatch [22-24]. The framework is highly adaptable; by swapping out the Apify scraper, it can generate content based on Reddit, Amazon, LinkedIn, or YouTube data [25, 26]. * **The AI Management Philosophy:** The overarching philosophy of the system is that AI should be treated as a worker, while the human acts as the manager who builds the system, provides real market data, and reviews the final output [27, 28]. ### TBM 412: Institutionalized Overload (Now With AI) by John Cutler * **Normalization of Overload:** Organizations have deeply normalized cognitive overload and constant work-in-progress, to the point where processing excessive context has become ingrained in professional identity [29, 30]. * **AI as an Amplifier, Not a Cure:** Instead of disrupting this broken system, AI is currently being used to help people cope with and sustain the chaos [29, 31]. AI allows workers to juggle more inputs, reinforcing the very conditions that make thoughtful work difficult [30, 32]. * **Internalized Pressure:** The pressure to handle more and respond faster becomes internalized, making focused, calm, and efficient work feel uncomfortable or incorrect [33]. * **Resisting Expansion:** As AI expands our capacity to process context, the amount of information and signals expands to fill it; a key part of modern work will involve actively resisting this maximalist expansion of tasks [32]. ### TBM 416: Investment Stewardship (As Habit) - by John Cutler by John Cutler * **ROI is a Habit, Not a Calculation:** Determining the return on engineering investments is not a simple math equation or dashboard metric, but a continuous set of behaviors rooted in stewardship and thriftiness [34-37]. * **The "Cold-Start" Problem:** Companies often fail to measure ROI because they wait until there is a crisis to ask the hard questions, trying to "cold-start" the evaluation after years of ignoring outcomes and celebrating mere output velocity [38-40]. * **Finding Good Proxies:** Because product work is often several steps removed from financial value, leaders must avoid the "precision or nothing" trap; relying on good-enough, directional proxies is far better than having false confidence from bad proxies like story points [41, 42]. * **Building the Stewardship Habit:** Achieving proper ROI tracking involves establishing leading indicators, building useful business models, consistently updating confidence through signals (Bayesian updating), and maintaining strict discipline regarding hiring and team funding [42-44]. ### The Convenience Trap - by Jessica Talisman by Jessica Talisman * **The AI Productivity Paradox:** Despite massive investments, macroeconomic data and major surveys (like a 2026 NBER study showing 89% of firms report no productivity impact) suggest AI is not delivering broad productivity gains [45]. * **The Verification Burden:** While AI generates content rapidly, workers lose significant time reviewing, editing, and verifying the output due to high hallucination and bug rates, sometimes resulting in a net loss of time [46-48]. * **Jevons Paradox and "Workslop":** Applying the Jevons paradox to knowledge work, the author notes that making content creation cheaper only increases the demand and output [49]. This has flooded the internet and workplaces with "workslop"—high volumes of low-quality, synthetic content that demands human attention [50-52]. * **The AI Efficiency Trap:** The surge in generated content leads to burnout as workers face an infinite supply of output needing review [53, 54]. Workers become dependent on tools (agency decay) while the organizational expectation for output continually increases [55]. * **The "Strip Mall" of Knowledge Work:** AI has made production cheap and accessible but downgraded its quality, generating an overabundance of content without saving humans time or reducing their workload [56, 57]. ### The data reliability question you're avoiding by Andrew Jones * **Speed Over Reliability:** Data engineers frequently prioritize moving quickly over ensuring data reliability, often relying on quick fixes, neglecting monitoring, and skipping investments in reliability tooling [58, 59]. * **Misaligned Expectations:** If users are expecting highly reliable data to power critical business processes or revenue-generating machine learning features, the "speed first" trade-off is inadequate [59]. * **The Needed Choice:** Engineers must either actively change their trade-offs to build more reliable pipelines or ensure that user expectations are clearly aligned with the reality of how the data is being delivered [59, 60].