RB2S Conseil — AI & information systems advisor
I advise company directors and IT departments on adopting artificial intelligence and on the infrastructure that makes it possible. Nineteen years of advisory work and information systems leadership, including six running the operations of an industrial SME that grew from twelve to forty-seven employees: I deployed these tools and carried the consequences before talking about them.
AI advisor
Almost every company has tried an assistant. Very few have a use that holds up over time, with data under control, a known cost and written rules. That is exactly where I work: between the enthusiasm of the trial and the demands of running it in production.
A review of candidate tasks, with an honest estimate of the time saved, the cost of use and the risk of error. We keep three, we drop the rest, and we know why.
Proprietary models over an API or open models run on your own premises, retrieval-augmented search, agents, connections to your business applications. A costed comparison, not a vendor preference.
Where your documents go, who keeps them, what is used for training, what stays in France. Sensitive data kept separate and, where necessary, local inference on your own hardware.
A usage policy, classification of systems under the European AI regulation, logging, human validation of decisions, and how it all fits with the GDPR and your sector obligations.
Workshops by function rather than generic training: we start from their own files, their document templates, their awkward cases. Teams leave with uses they can handle themselves.
Indicators before and after, a consolidated monthly cost, and a decision to extend or to stop. An AI project that is not measured becomes a cost nobody dares question.
Better said straight away
A proof of concept with no route to production is money spent on appearances. If the use case will not survive the next stage, I say so at scoping, not at the review.
I do not resell licences, platforms or API credits, and I earn nothing on your choices. It is the only way to make a decision between two suppliers credible.
AI shifts tasks; it does not do away with jobs overnight. The gains I state are expressed in hours recovered on specific, verifiable tasks.
Information systems and infrastructure
Scattered data, a fragile network or approximate security will ruin the best AI project. These four assignments deal with the foundations, and work just as well on their own.
Where your information system actually stands, what has to be dealt with this year, what can wait, and at what cost. A budgeted roadmap a board can make decisions on.
Hosting choices, public cloud, private cloud or your own machine room, multi-site networks, data location and exposure to extraterritorial laws. An architecture paper and a costed comparison.
The real level of risk, prioritised work, business continuity and recovery planning, GDPR and NIS 2-type requirements, and preparation for the audits your clients and insurers run.
Part-time IT leadership, taking back control of a project that is drifting, IT due diligence before an acquisition and a systems integration plan after the deal.
How it runs
An hour by phone or video call to understand the context and check that the subject really is in my field. If it is not, I say so and point you elsewhere.
No chargeA day on site with your business teams: an inventory of possible uses, sorted by value and by risk, and a first architecture. You leave with a document you can use on your own.
The most requested way inInterviews, analysis of what exists, technical trials where useful, then a written report and a presentation to the board.
5 to 20 days depending on scopeSupport through implementation if you want it: steering, review meetings, decisions, measurement of the gains. No obligation and no commission on the suppliers chosen.
Fixed price or time and materialsAssignments in French and English, remotely, or on site in the Grand Est region and in neighbouring Germany. Day rate given at scoping to be confirmed.
Background
Out of respect for my former employers and my clients, I describe the contexts rather than the names. Precise references are given verbally, on request, at the first conversation.
RB2S Conseil
Supporting company directors and IT departments in adopting AI: strategy, architecture, data confidentiality, governance and change management.
Industrial SME in the Grand Est region
Growth from twelve to forty-seven employees and three branches opened. ISO 9001, 14001 and 45001 management systems deployed through to certification. Digital tools put in place — customer relationship, human resources, purchasing — along with the first uses of intelligent automation. Cybersecurity and GDPR compliance strengthened. Negotiated and completed the sale of the company to investors, then regional director through the transition.
International industrial group — Paris La Défense
Running multi-country IT programmes in a heavily regulated industrial environment.
Enterprise software publisher — Paris region
IT service management and operations automation for large accounts.
Technology consultancy — Paris
First advisory assignments on infrastructure and on the organisation of IT departments.
School of computer engineering
And afterwards
A lot of recommendations come to nothing for want of people to apply them. For organisations in Centre Alsace, our experienced staff take over day-to-day operations — including deploying AI assistants in the tools your teams already use. And if your plans require the processing to stay in France, we have our own machine room, able to host inference machines.
Contact
I reply personally within twenty-four working hours, and the first conversation is not charged for.