Find the decision worth testing
A short diagnostic on one process, one data family, and named owners. You get a go / no-go and a statement of work. A no-go is a successful result if the current tool already works.
Controlled AI workflows. Stop if the evidence is weak.
NEXUS works with leadership teams who need a decision they can defend - what to change, what to leave alone, and how to test it.
We lead discovery, design and bounded delivery on your approved data. Where production engineering or sector assurance is required, we contract named partners. We do not sell chatbot access as a strategy.
A typical first step is a short diagnostic on one process. If the evidence is weak, we stop. If it is strong, we design a read-only working method with written tests and a human owner.
The first conversation is a scoped engagement, not a course and not an assessment product.
This is how the work is owned
The problem is not understanding a model. It is choosing a process that can be tested, naming an owner, and stopping when the evidence is weak. That is the team you get with NEXUS AI Advisory.
We design controlled workflows on approved records. Rules first, models only where they beat the baseline. Outputs are proposals. Your system of record stays in charge.
Dr Ioannis Syrigos owns the engagement: scope, terms and when to stop. Athanasios Serafeim owns commercial delivery and whether an offer can be executed on the ground. Prof Nikos Fakotakis owns method and research quality. Lucie Loiseau owns data governance and ethics, including the Know Your Data (KYD) method. Dimitris Vrachnos owns adoption and whether the change survives the operating model. Partners are named when the work needs production engineering or sector assurance.
The first invoice is a scoped test and not a transformation programme.
Build a read-only working method
Opportunity cards, evidence registers or review queues linked to source records. No automatic purchase order, supplier award or bid submission. Fees and tests are fixed before build.
Hand over, scale, or stop
If quality and ownership hold, we plan production inside your approved environment. If they do not, we stop. We do not sell a transformation programme as the first invoice.
Method
We do not start with a model. We start with the decision you have to stand behind, the records you are allowed to use, and a test that can fail.
Rules and source records come first. A model is added only where it beats that baseline on quality and effort.
Every output is a proposal. Your system of record and a named owner stay in charge.
Live data is used only after purpose, classification and environment are approved. Permission for one tool is not permission for another.
The durable asset is the workflow: matching rules, evidence lineage and acceptance tests and not a brand of model.
Dr Syrigos I.
CEO
Engagement owner. Engineering, AI and management background; lecturer and company leader. Sets scope, commercial terms and when to stop.
Prof Fakotakis N.
VP of Operations
Method and research quality. AI and natural language processing; 40+ European and Greek R&D projects; 400+ scientific publications; co-founder of an early Greek AI company.
Vrachnos D.
Strategic Growth & Business Development Advisor
Adoption and whether a change survives the operating model. 25+ years in commercial transformation, including leadership roles at Allergan and L’Oréal.
Serafeim A.
Partner & Chief Commercial Officer
Makes sure an offer can be delivered. Entrepreneur and managing director with two decades in operational and commercial leadership, including digital transformation in technical production and hospitality. Owns client commercial relationships.
Loiseau L.
Data Governance & AI Ethics Advisor
Data rights, quality and accountability before a model is introduced. Advises on ethical AI and governance in regulated European settings. Know Your Data (KYD) method.
