SuperSimple.nl

Making AI useful

SuperSimple.nl focuses on applied AI: not a standalone chatbot demonstration, but solutions connected to documents, databases, APIs and daily work. The technology should remain understandable and deliver measurable value.

My AI experience

As an AI Architect & Lead Developer, I work on AI products from the first use case and architecture through prototype, integrations, production and further development. I am comfortable with modern AI tools and turn new capabilities into workable solutions quickly.

LLMs and AI agents

• ✓Integrating large language models into web applications and business processes.
• ✓Designing AI agents that use tools, APIs and data sources in a controlled way.
• ✓Working with prompts, system instructions, structured output and tool calling.
• ✓Orchestrating multi-step workflows with clear boundaries and checkpoints.
• ✓Evaluating responses for reliability, source use and practical usefulness.

RAG and trustworthy knowledge

Retrieval-Augmented Generation retrieves relevant documents and knowledge sources before a model answers. This creates contextual answers with citations instead of a generic chatbot without visible supporting evidence.

MCP, databases and integrations

• ✓MCP connectors that let models communicate safely with external tools and systems.
• ✓Vector search with Qdrant and Pinecone for semantic retrieval.
• ✓Connections to PostgreSQL, MySQL, MariaDB, APIs and internal data sources.
• ✓Read-only and controlled access where data safety requires it.
• ✓Laravel, PHP, Python, React and GitLab CI/CD as a production foundation.

AI products in practice

The experience includes Huurrecht AI, CRA Operations and AI workflows around document search, release evidence and business data. The focus is always the combination of good source data, a clear user flow and a system that remains maintainable.

Comfortable with AI tools

I use AI tools as part of analysis, development, documentation and product design. I also look critically at model choice, cost, privacy, failure modes and whether automation genuinely helps.