Databases, automated workflows, AI models, business interfaces, hosting, security, regulatory compliance, and API cost management: an AI initiative quickly evolves into a full-scale digital architecture project.
It is precisely to explore and master these operational challenges that OIDANEOS is progressively building its own internal AI experimentation platform.
Moving from Using AI to Engineering AI Solutions
Today, accessing artificial intelligence is remarkably easy. It takes only a few seconds to open ChatGPT, Claude, Gemini, or Mistral.
However, when an organization wants to connect AI to its own proprietary data or automate a critical business process, the engineering questions become far more complex:
- Where is the data stored?
- How should it be structured and governed?
- Which AI model is best suited for the task?
- How do we orchestrate and connect disparate business applications?
- Where should the solution be hosted?
- How do we secure access and enforce permissions?
- How do we control and optimize API consumption costs?
- How do we ensure full compliance with GDPR and the new EU AI Act requirements?
At this point, AI ceases to be a simple conversational assistant and becomes mission-critical business infrastructure.
02 - ExperimentCreating a Dedicated Environment to Test AI Technologies
OIDANEOS has established a dedicated technical environment designed to host, benchmark, and test a wide range of technological building blocks. The architecture relies on dedicated server infrastructure and self-hosted open-source solutions.
Data & Applications
Baserow for structured data management, Celestory for interactive interfaces and workflows, Moodle for learning environments, and a suite of open-source collaborative tools.
Controlled Infrastructure
Containerized deployment using Docker to test multiple isolated architectures in parallel and scale solutions progressively.
Sovereignty & Autonomy
Rather than relying systematically on external SaaS platforms, core building blocks are engineered and operated within a fully controlled environment.
Data as the Foundation
One of our primary areas of experimentation involves relational databases designed around real-world enterprise use cases. Many of these comprise interconnected tables that already mirror the architecture of a future business application.
Our objective is to ingest this data structure, rebuild it within the OIDANEOS environment, and evaluate:
- Data relationships and schema integrity;
- Potential optimizations to the relational model;
- User interfaces tailored to operational teams;
- End-to-end workflow automation opportunities;
- How AI systems can securely interact with this information (RAG architectures, autonomous agents).
Rapidly Developing Business Applications
Generative AI is also fundamentally reshaping software engineering. In our lab experiments, a user interface can be conceptualized, wireframed, and rapidly transformed into a functional application. The underlying code can then be refined and hardened with the assistance of AI coding agents.
Why this matters: Prototyping a custom business application no longer requires launching a heavy, multi-month IT project upfront. Organizations can start small—a structured database, a clean interface, and an initial automated workflow—and iterate progressively.
05 - Stay AgileBenchmarking Multiple AI Models
Building an independent architecture also prevents vendor lock-in. Depending on the use case, an application can dynamically route tasks across models from OpenAI, Anthropic, or Mistral, or leverage open-weights models running in dedicated private environments.
This model-agnostic agility is a strategic advantage. Today's leading model may not be tomorrow's optimal choice. The challenge is therefore not merely selecting an AI provider, but engineering an architecture capable of evolving alongside the technology.
Master AI Solution Architecture
OIDANEOS trains and supports IT professionals in designing secure, high-impact AI solutions tailored to their business operations.
Embedding Security by Design
Greater technical autonomy brings heightened responsibility. During testing and prototyping phases, live customer data must never be transmitted unprotected to external AI services. Schemas are replicated and datasets are anonymized so teams can experiment without exposing sensitive information.
Strict access controls and API rate governance are equally essential. A misconfigured application can easily trigger runaway API requests and incur significant token consumption costs.
A Mission That Extends Beyond Training
This experimentation platform also reflects how we view our mission at OIDANEOS. Training professionals on AI tools remains essential, but organizations increasingly need to go further: understanding how to integrate these technologies into their core systems, data pipelines, and business processes.
Our internal platform enables us to stress-test these architectures before translating them into real-world enterprise deployments:
- Technical Lab: Benchmarking technologies, foundation models, and system architectures.
- Pedagogical Environment: Demonstrating in practice how every architectural layer works together.
- Prototyping Studio: Progressively turning operational use cases into production-ready business applications.
Experimenting Today to Build Tomorrow's AI Infrastructure
Artificial intelligence is evolving at an unprecedented pace. Searching for a static, permanent architecture today would be an illusion. Our methodology is deliberately pragmatic:
Test. Measure. Understand. Improve.
Some tools will fade away; others will become industry standards. AI models will continue to advance. Yet one trajectory is already clear: enterprises will not rely indefinitely on generic, off-the-shelf chat interfaces.
They will progressively build their own sovereign environments, securely connected to their proprietary data and tailored to their operational realities. That is the transition OIDANEOS is actively exploring and championing today.


