Every weekday briefing, archived. Search by keyword or filter by signal to find the issue you remember.
The recent cadence of AI announcements shows a clear shift in how technology vendors frame their capabilities. The industry conversation is moving away from sheer parameter counts and toward context retention, operational latency, and domain adaptability. For those of us responsible for enterpri…
Read issue →Board requests for enterprise AI strategies rarely arrive with reasonable timelines. Most executive teams want to see immediate progress, which often leads IT departments to sponsor dozen of uncoordinated software pilots that never reach operational maturity. A structured 90-day execution window…
Read issue →For decades, we treated technical debt as an uncomfortable tax on IT velocity. We knew legacy code, hardcoded integration scripts, and undocumented middleware were slowing down software updates, but as long as core systems processed transactions, boards accepted the drag. The rise of enterprise…
Read issue →Most corporate reskilling programs fail because they focus on generic awareness rather than mechanical role shifts. Buying your engineering department a subscription to an online learning platform and calling it a reskilling initiative is theatre. It yields high completion certificates and zero c…
Read issue →In the rush to adopt generative models, leadership teams often overlook traditional, deterministic automation. Yet roughly sixty percent of standard corporate workflows—data validation, scheduled reporting, standard routing—require absolute predictability, not probabilistic reasoning. When you d…
Read issue →Executive teams face persistent pressure to deploy artificial intelligence across operations, which frequently leads to binary debates. Teams either rush to sign vendor contracts for immediate functionality or insist on building custom solutions to preserve control. Neither extreme serves the bus…
Read issue →A board member asked me recently if pasting a draft P&L into ChatGPT meant OpenAI's competitors could see it. The quick answer is no—vendors like OpenAI, Anthropic, and Google do not trade data with one another. The longer answer requires nuance. What you upload to a free or standard commercial…
Read issue →Over my decades in enterprise IT, I have seen too many executive teams freeze while trying to build the definitive master plan for new technology. Generative AI is creating that same paralysis today. Boards demand a transformative vision, but operating leaders simply need to fix chart backlog, lo…
Read issue →Every technology cycle brings a new wave of vendor arithmetic. We saw it with client-server in the nineties, with enterprise resource planning a decade later, and with cloud migration after that. Now, boards are asking the same simple question about generative AI investments: where is the net dol…
Read issue →As we deploy large language models and specialized runtimes into production, we are introducing two primary entry points into our enterprise boundaries: direct manipulation through prompt injection, and untrusted code via the model supply chain. Neither threat requires sophisticated cryptographi…
Read issue →Every board wants to hear about the next generative model or predictive pilot. Few want to discuss data lineage, contracts, or master records. Yet in forty years of managing enterprise systems, I have never seen an advanced capability overcome an unreliable data foundation. If your underlying da…
Read issue →For forty years, I have watched governance committees turn good intentions into bureaucratic gridlock. When new capabilities arrive—whether it was relational databases in the 1980s or generative models today—the instinct of risk management is to erect a tollbooth. The problem is that tollbooths…
Read issue →The vendor demonstration usually shows an agentic AI system effortlessly resolving complex customer inquiries, updating financial ledgers, and adjusting inventory levels without human intervention. In the boardroom, this looks like pure efficiency. In the executive suite of an operating business,…
Read issue →When cloud migration started fifteen years ago, many IT organizations made the mistake of tracking server hours instead of business workload outcomes. We are making the exact same error with generative AI today by focusing on cost per token rather than cost per resolved task. Inference costs rar…
Read issue →Most enterprise software projects fail due to poor scope, but AI initiatives typically stall because nobody knows who actually owns the operational outcome once a model goes into production. When a board asks who is responsible for an automated decision, the usual answer is a committee rather tha…
Read issue →Enterprise AI initiatives rarely fail because the underlying math was flawed. They die quietly in sandbox environments because nobody prepared the operational environment for what happens on day two. Over four decades in IT, I have watched this exact cycle repeat with expert systems, data lakes,…
Read issue →Board members and executive committees are asking a familiar question this quarter: have we missed the boat on AI? Watching competitors issue press releases about generative tools creates understandable anxiety at the leadership table. Having managed enterprise IT transitions since 1980, I can a…
Read issue →When a solar eclipse occurs, the physical effects on server hardware are minimal, but the indirect operational ripples are real. Solar power generation drops rapidly across affected regions, cell towers choke on localized crowd density, and subtle changes hit the upper atmosphere. In my four dec…
Read issue →Two founders launching a business in rural West Virginia face classic operational constraints: tight capital, a limited local labor pool, and reliance on strong personal relationships. Over my forty years in IT, I have watched small operators try to solve these problems by working longer hours un…
Read issue →Executive schedules in multi-site businesses are consumed by operational fires across dozens or hundreds of locations. Finding structured hours to study artificial intelligence feels near impossible when you are managing day-to-day outages, staffing shortages, and vendor integrations. Yet, falli…
Read issue →When major technology firms publish demonstrations of new AI hacks, board members usually take notice. The headlines focus on theoretical worst-case scenarios, leaving executive teams wondering if their own deployment of artificial intelligence is an open door for attackers. Panic is never a str…
Read issue →Boardrooms are asking harder questions about the sheer volume of capital earmarked for artificial intelligence. Over four decades in this industry, I have watched several cycle peaks where technology spending detached temporarily from near-term economic utility. Whether market commentators call i…
Read issue →For decades, enterprise software budgeting relied on predictable seat licenses. You counted user accounts, negotiated a tier discount, and set the baseline budget for the fiscal year. Generative AI has upended that stability. Vendors are rapidly shifting to variable consumption pricing—usage cre…
Read issue →In my forty years of infrastructure and software rollouts, I have watched the same mistake repeat itself: leadership mistakes a vendor's polished sandbox for proof of operational readiness. With generative models and clinical decision tools, the pressure to deploy quickly across thirty or three h…
Read issue →Social media feeds and vendor marketing campaigns are currently saturated with claims that every software updates powered by artificial intelligence will fundamentally transform your balance sheet. For executives who have managed technology cycles through the client-server era, the dot-com buildo…
Read issue →In my four decades leading IT organizations, I have seen every technology cycle produce its own version of the pilot graveyard. In the 1990s it was client-server migrations; a decade ago it was unstructured data lakes. Today, generative AI initiatives are stalling at record rates, usually right a…
Read issue →Most dental organizations don't have a demand problem — they have an execution problem. A special edition on why operational discipline, not lead generation, defines the next decade of dental growth.
Read issue →Special edition. Healthcare digital transformation fails at a higher rate than almost any other industry — not because the technology is harder, but because the operating model is. This is the playbook I ran across six CIO seats, from hospital-adjacent systems to multi-site dental and veterinary…
Read issue →Good morning. The AI buildout got its first real bill this week — and the market flinched. Plus: OpenAI's agent broke containment, dental AI hits table stakes, and the veterinary PIMS race goes AI-native.
Read issue →