From board-ready AI strategy to working systems — designed and built by the same principal. I assess where AI creates measurable value in your specific operation (grounded in task-level operating data), your data readiness and governance, and the build/buy/partner economics — then build it: retrieval over enterprise data, multi-agent workflows, ML classification and scoring, and AI embedded in ERP-integrated applications. Grounded in a production stack operated hands-on in this practice, not vendor slideware.
The Problem
Every vendor is selling you AI. Almost none will say the uncomfortable truth: your AI roadmap is only as good as the data underneath it — and in most mid-market operations, the data is not ready.
Fragmented systems, inconsistent master data, undocumented processes, tribal knowledge
AI initiatives stall — or worse, automate the wrong things confidently
Nobody tells you which problems a language model should never own — numeric forecasting, for one
The gap between boardroom AI enthusiasm and P&L AI value keeps widening
The Engagement
Most AI advisory stops at the slide deck. This lane runs from strategy through working software, delivered by a practitioner who builds and operates production AI systems — retrieval pipelines, multi-agent orchestration, ML classification — with the same environment, testing, and data-governance discipline applied to enterprise platforms.
AI Value Creation Assessment (packaged entry point): A structured assessment of where AI creates measurable value in your specific operation — use cases grounded in task-level operating data, governance as a first-class design input, build/buy/partner economics, and a prioritized roadmap tied to operating impact
AI-ready systems and data: The unglamorous work that makes AI possible: data quality, master data governance, integration architecture, and migration discipline drawn from 100+ enterprise systems engagements
Design and build: RAG and knowledge systems over enterprise data; multi-agent workflows with human checkpoints; ML classification, scoring, and predictive models; AI embedded in ERP-integrated applications (Oracle APEX AI on JDE data)
Governance by design: Pseudonymization upstream of every model call, evaluation and quality gates, human oversight, and cost controls as design properties — not afterthoughts
The strategy and the build are the same engagement and the same principal: the architecture recommended is the architecture delivered.
Proof
PE-backed hospital and physician multimedia company (Interim CTO & Independent Board Member): R-based ML profiling and scoring across 260M+ records (InfoUSA, hospital EHR, NHANES) powering targeted variable-print and precision marketing campaigns; stood up the IT and Data Science departments that took development in-house
Plastic molding job shop: ML data mining across millions of historical production records disproved the “no commonality” hypothesis — 5 core production profiles generated as the basis for a CPQ application; future-state Quote-to-Cash and PLM approved by C-level leadership
Distribution & fulfillment plant: ML text-matching MVP for automated Amazon A/R reconciliation, designed to eliminate over 90% of a manual process spanning 18+ databases and 60+ spreadsheets
Enterprise-system selection at scale (PARA™ RFI framework): Embedding-based semantic requirement matching with human checkpoints, transformer sentiment analysis of vendor-demo feedback, and weighted must-have coverage scoring — applied across five selection engagements and 880+ weighted requirements
Practice-built, practice-operated AI systems: Debrief — a governed LangGraph multi-agent meeting-to-deliverables application (FastAPI, PostgreSQL/pgvector, Docker) with pseudonymization at ingest, leak scan with quarantine, a token-budget guard, and a per-call run log — measured at 47–72 seconds end to end and roughly $0.06 per run with zero leaks on its synthetic golden set; and CaseForge, the NER-based anonymization gate every downstream AI call passes through. Built within the practice and applied to engagement delivery.
Enterprise AI Strategy, Solutions & Hands-On Development
Where AI creates measurable value in your operations — and where it does not.
