Martin Taal

Technical & Functional Software Architect — AI × ERP

Independent consultant with deep experience at the intersection of ERP architecture, model-driven development, and agentic AI systems. Serving international clients across all regions and time zones from the Netherlands.

This practice covers the full spectrum of business software and ERP: from ground-level data modelling and API design to customer-facing solution architecture and pre-sales. Architecture decisions are grounded in real business requirements, not technology for its own sake.

Current focus is at the crossroads of AI and ERP — integrating agentic systems with deterministic ERP engines, and shaping ERP data models into representations that are natively consumable by AI systems.

A commitment to open source development is shown in past work for the Eclipse Foundation as a committer and project lead of the EMF Teneo and Texo projects.

A unique combination of technical depth and functional breadth defines this practice. With extensive experience in the business software and ERP space, work spans ground-level data modelling and API design all the way to customer-facing solution architecture and pre-sales.

Both the technical and the functional architecture perspectives are covered — designing systems that are robust, maintainable, and grounded in real business requirements rather than technology for its own sake.

Clients include independent software vendors, system integrators, and enterprise IT departments who need a trusted sparring partner who understands ERP processes end-to-end: from procurement and contract management to inventory, rental, and financial modules.

Capabilities

• ERP system architecture — technical and functional
• Business process modelling and requirements analysis
• Contract, rental, inventory, and procurement domain design
• Pre-sales support and customer-facing solution presentation
• Model-driven development with EMF, JSON Schema, XML Schema
• REST API and web-service architecture
• Continuous delivery: test-driven, iterative, short feedback cycles
• Java stack: Hibernate, JPA, OSGi, Tomcat, PostgreSQL, Oracle, MSSQL
• Rich JavaScript front-ends integrated with ERP backends
• Proof-of-concept and prototype development

The most interesting challenge in enterprise software today is understanding exactly where agentic AI and large language models add value — and where deterministic ERP logic must remain in control. Getting that boundary right is an architectural question, not just a technology one.

Agentic Systems Integrated with ERP

Agentic AI systems — autonomous software agents that plan, act, and adapt — can dramatically accelerate ERP workflows when wired correctly. The focus here is on connecting LLM reasoning layers to deterministic ERP transaction engines: the AI decides what to do; the ERP ensures how it is done with full auditability, consistency, and business-rule enforcement.

Practical work includes designing tool-calling interfaces between LLM agents and ERP APIs, structuring agentic workflows for procurement, contract lifecycle, and inventory management, and ensuring that ERP state remains the single source of truth throughout.

Modelling ERP Data Structures for AI

ERP data models are complex, relation-heavy, and domain-specific. Making them consumable by LLMs requires a deliberate modelling layer: extracting semantic intent from schema definitions, representing ERP entities as structured context for retrieval-augmented generation, and building datasets grounded in real business ontologies.

The structural discipline of formal metamodelling — developed over years of Eclipse Modeling Framework work — translates directly into producing well-shaped, AI-ready representations of ERP domains.

A long-term commitment to the Eclipse open source ecosystem as an active committer and project lead produced two widely adopted frameworks in the Eclipse Modeling Framework (EMF) space.

Eclipse Modeling Framework (EMF) is a meta-modelling and code generation platform that was central to this work, providing the formal structural foundation for model-driven ERP and web-application development.

The model-driven discipline developed through this Eclipse work directly informs the approach to shaping ERP data models for AI systems: formal, rigorous representations that both machines and domain experts can reason about.

Springsite is the independent consultancy company of Martin Taal, a technical and functional software architect with three decades of experience in business software, ERP, and model-driven development. The practice operates remote-first and serves clients across different regions and time zones.

Core strengths lie at the intersection of deep technical knowledge and practical business understanding: designing systems that solve real problems without unnecessary complexity, and communicating architecture decisions to both technical and non-technical stakeholders.