Kevin Clark
My focus is AI strategy, agent development, evaluation, and the organizational work that turns a working model into a useful system.
Read selected work ↓// 01 — ABOUT
I work on the practical questions of AI: what to build, how to evaluate it, and how to make it useful in a real workflow.
My focus spans AI strategy, agents, automation and governance. I care about the decisions around a system as much as the model inside it: where human judgment belongs, what quality means, and what happens when the system fails.
This site brings together independent technical projects and writing about AI deployment, evaluation and organizational adoption.
// 02 — AI PRACTICE
Choose a workflow with a clear purpose and an owner. Understand the decisions, inputs and constraints before deciding where AI belongs.
Read about deployment →Define what good enough means for the task. Test against a baseline, account for human review, and use the results to decide whether to expand, revise or stop.
Read about evaluation →The projects below explore systems and the infrastructure behind AI. The writing examines how strategy, evaluation, governance and adoption shape what those systems can do.
Explore independent projects ↓// 03 — EXPERTISE
01
Workflow priorities, technology investment, vendor evaluation, and build-versus-buy decisions.
02
Task design, tool use, integrations, and human review for AI-enabled workflows.
03
Task-specific quality criteria, baselines, failure analysis, and deployment thresholds.
04
Workflow redesign, practical training, feedback, and clear ownership after launch.
05
Data handling, oversight, accountability, and the limits of automated decisions.
06
Exploring the power, water, and physical systems behind AI through independent research.
// 04 — DEPLOYMENT
From workflow discovery to sustained adoption. Each phase needs clear ownership and evidence.
// 05 — INDEPENDENT PROJECTS
Independent technical projects exploring automation, decision systems, and the infrastructure behind AI.
A market-monitoring project with an interactive strategy simulator. Adjust signal and risk parameters to explore behavior on synthetic market data.
A public research atlas of US data centers, generation, historical grid regions, reported water use, and watershed context. Sources and verification limits accompany the data.
// 06 — WRITING
Arguments about what actually works. Written from inside the problem, not above it. Follow on LinkedIn for shorter takes between essays.
Why a working AI prototype can stall between testing and everyday organizational use.
How baselines, success criteria, and workflow-specific evaluation help organizations decide which AI systems to adopt.
Why strategies that work at AI-first companies create debt when applied inside traditional enterprise, and what actually works instead.
// 07 — PERSPECTIVE
These principles inform my work on enterprise AI.
Model capability matters, but so do the decisions the system supports, the review it requires, and the feedback it creates.
A tool can work in testing and still fail in daily use. Workflow design, trust, and practical training need as much attention as the integration.
Good enough for what, compared to what, according to whom, at what cost? These are investment decisions as well as technical questions.
Changing an employee workflow consumes attention and trust. The pace of change needs to account for both.
Governance, evaluation, and reliable systems make strategic choices usable. They belong in the strategy from the beginning.
// 08 — CONNECT
For conversations about AI strategy, agents, evaluation, or technical projects: