Browse Dr. Joe Shepherd's writing on AI governance, PE operations, board advisory, and product leadership.
Browse the archive in order, move by topic, or jump to search when you already know the operating question you want answered.
Showing 12 of 18 articles · Page 1 of 2
Agentic systems are the engine. Discovery is the output. Once you separate the two, the strategic picture for operators, boards, and capital allocators looks very different.
Read →AI is already making decisions at scale in most PE-backed and regulated businesses. The question is not whether to allow it — it is whether you designed it.
Read →As AI agents absorb more search, evaluation, and purchase work, brand shifts from narrative alone to machine-readable trust. Operators need a playbook for winning both human preference and agent recommendation.
Read →AI oversight in 2026 is no longer a technology update. Boards and PE sponsors need a defensible, evidence-based operating model that regulators, buyers, and insurers will recognize.
Read →Most private equity firms still talk about AI as a set of use cases. The firms that will create real value will treat it as a portfolio operating system spanning thesis, governance, execution, and exit proof.
Read →AI product teams fail when leaders treat data science and software engineering as the same workflow with different job titles. The more effective approach is to recognize the overlap, keep the lifecycle stages distinct, and switch the team’s operating mode at the right moment.
Read →My doctoral dissertation examined how U.S. defense vendor executives approach cloud-modern solutions under ATO constraints. The findings were less about compliance theater and more about expertise, funding, reuse, data governance, and operability.
Read →Defense transformation fails when programs stop at software factories, landing zones, and cloud migration mechanics. Real progress requires a mission-led operating framework that connects people, process, technology, and battlefield outcomes.
Read →Most bloated backlogs are really avoidance devices. The three sprint rule forces teams to keep only a short window of real commitment, scope MVPs to fit that window, and let fresh customer signal reset the priorities.
Read →The wrong way to assign AI leadership is by defaulting everything to data science or everything to engineering. The right way is to ask whether the team is still exploring the path forward or already implementing against a clear one.
Read →Most AI board reports describe enthusiasm and pilots. This template forces management to show how AI labor is changing operating metrics, who owns the outcomes, and what decisions the board now needs to make.
Read →When Microsoft licensing shifted to monthly SaaS, our revenue line fell off a cliff. We rebuilt a $1.2B business by redesigning how we attached services to software. The same structural break is happening again as AI compresses SaaS economics.
Read →