Three decades of production business applications that close the gap between what the ERP does out of the box and what the operating model requires — 53 documented applications, nearly every one architected, built, integrated, and delivered by the principal who diagnosed the need. The stack has evolved (RPG, .NET/SQL Server, Oracle APEX/PL/SQL, Python); the discipline has not. Today that includes AI-assisted design and development, compressing concept-to-production without loosening the engineering standards.
The Problem
Packaged software should carry as much of the business as possible. But some processes are too differentiated, too cross-functional, or too important to force into a generic product. That is where spreadsheets multiply, manual workarounds become permanent, and critical knowledge ends up trapped in a few people.
ERP and SaaS platforms cover the standard process, not the operating edge that actually differentiates the business
Point solutions and spreadsheets accumulate until nobody owns the end-to-end workflow or the data moving between systems
Development vendors build what was specified without asking whether the underlying process should exist — and hand off without the training, environments, or data structures to run it
Agentic coding has collapsed the cost of writing custom software. It has not collapsed the cost of owning it. The build decision deserves more discipline now, not less.
The Engagement
The sequence is fixed: consult, then optimize the assets you already own, then implement commercial solutions, and only then build — each option exhausted before the next. When a build is justified, the same principal who diagnosed the gap architects and delivers the system, data structures included: no hand-off, no translation loss, no separate SI to manage. Four documented exceptions where I architected and a team I hired or directed built — I say so unprompted.
ERP-integrated custom applications: Workflow, planning, decision-support, and operational tools that leverage and extend the master, transactional, and historical data your enterprise systems already govern
Bolt-ons, middleware, and integration: Interfaces, staging and reconciliation layers, and the data-generation programs that create the master data packaged systems lack
Algorithm and analytics engines: Forecasting, ABC classification, costing, reconciliation, capacity modeling — the math ERPs approximate
Oracle APEX / PL/SQL on JDE and Oracle-schema ERPs: Custom UI, reporting, and AI layers directly on the ERP database
Delivery discipline: Environment lifecycle and CI/CD, testing gates, change-audit and run-log tables, code review and power-user validation, stakeholder prototyping, training, and knowledge transfer so your team owns what gets built
AI-assisted SDLC and legacy modernization: Concept-to-production compressed with AI-assisted design and development, held to the same evaluation, governance, and cost-control gates
Proof
Global 500 turbine blade & generator wedge division: Heuristics-based configurator/estimator integrated with JDE — blade quotes in under 2 hours against a ~3-week industry standard; an emergency multimillion-dollar utility RFP estimated in half a day while competitors quoted two weeks or more — and won
National pipe distributor (250K SKUs, 18 DCs): PL/SQL SIOP algorithm suite on JDE EnterpriseOne — demand distribution, safety stock, ABC ranking with monthly recategorization, inter-DC transfers — running hourly through monthly; ~99% of manual planning and purchasing execution eliminated
PE-backed high-volume molding manufacturer: Oracle APEX process & discrete manufacturing application, core built by one principal in 4.5 months with a 10-minute polling integration from Dynamics GP; over 97% of data-entry errors eliminated, material shortages and the chargeback crisis resolved
$2B+ AUM investment manager: C#/.NET bolt-on suite on the LedgeX portfolio platform — data-integrity rule engine, NAV and Modified Dietz performance, time-driven ABC profitability, manager due-diligence middleware, and the investor-relationship module expanded into a full CRM; a 20-year Access/Excel estate migrated; ~95% data-quality improvement; 42 report sets consolidated to one
Distribution & fulfillment plant (SAP/WMS): Oracle APEX A/R reconciliation application with ML text matching for Amazon payments and chargebacks — future state eliminating over 90% of manual processing across 18+ Access databases and 60+ Excel workbooks; MVP delivered
Team lineage: Two developers stood up and mentored at an endowment investment firm went on to found Solovis — portfolio management software acquired by NASDAQ in March 2020
Enterprise Software Engineering
Custom applications integrated with, bolted onto, and embedded in your enterprise systems — designed and built by the same principal.
