From reception to the director's office: the journey that puts your AI in order.

Eight rooms, eight stages. We come in through reception, uncovering all your AI —including the hidden kind— and walk through your office until it's secure, compliant and led. Scroll: the office comes along with you.

CYBERSECURITYDATA GOVERNANCEAI
0TOOLS DETECTED
0HIDDEN FROM IT
0MAX AI ACT FINE

We walk your office, room by room.

Each room is a stage with a fixed price and its plain-language translation. The office lights up as you go.

STOP 01 · RECEPTION · 2–3 WEEKS

Diagnosis: the picture of all your AI

In plain English: we find out what AI is in your house —including the hidden kind—, who uses it, with what data and what it really costs.

  • Anonymous survey across the whole team: this surfaces the AI nobody declares.
  • Technical review: invoices, access and connected permissions.
  • Report with roadmap: the route map for the rest of the journey.

EXAMPLE · THE LIVE SCAN

SCANNING
CORPORATE STACK
ChatGPT — personal accountsSALES · FREE
Customer data
Copilot 365WHOLE COMPANY
Approved
Fireflies.aiRECORDS MEETINGS
No DPA
Notion AINOT DECLARED TO IT
Shadow AI
STOP 02 · LEGAL · 2–3 WEEKS § GDPR IN FORCE

Compliance: getting you in order

In plain English: we go through the paperwork. Making sure every tool meets the EU AI law and has a signed data agreement.

  • We classify each system by its AI Act risk level.
  • We review the data processing agreements (DPAs) with every provider.
  • Prioritized remediation plan, with a legal partner.

EXAMPLE · WHAT SHOWS UP

EXTRACT · DATA AGREEMENTS (DPAs)
DPAs reviewed and in order12
No signed DPA4 — to remediate
Data outside the EU without safeguards2 — to remediate
High-risk systems (HR)2 — obligations Aug 2026
STOP 03 · CLASSROOM · FROM 1 WEEK § MANDATORY SINCE FEB 2025

Training: your team up to speed (and compliant)

In plain English: we teach your people to use AI well and without risk. EU law already requires it of every company that uses it.

  • Program by role: leadership, managers and teams.
  • With your real cases, not generic theory.
  • Documented record: the evidence Article 4 requires.

EXAMPLE · WHAT SHOWS UP

EXTRACT · ARTICLE 4 PROGRAM IN PROGRESS
Leadership · 2 hCompleted
Managers · 4 hIn progress
Operations staff · 6 hWeek 3
Staff trained and on record64%
STOP 04 · BOARDROOM · 6 WEEKS § PREP FOR AUG 2026

Governance: the rules of the game

In plain English: we put the rules in writing: which tools can be used, by whom and with what data. And who decides when a new one comes along.

  • Usage policy of 4–5 pages: if it isn't read, it doesn't exist.
  • Matrix of approved tools, with an alternative for each banned one.
  • AI committee that decides in days, not months.

EXAMPLE · WHAT SHOWS UP

EXTRACT · TOOL MATRIX
ChatGPT Team
WHOLE COMPANY
Approved
Notion AI
NO CUSTOMER DATA
Conditional
ChatGPT free
ALTERNATIVE: TEAM
Banned
STOP 05 · STOCKROOM · 2–4 WEEKS

Consolidation: fewer boxes, better sorted

In plain English: we throw out the duplicates, keep what's useful and move it to company accounts. And on top of that, you save from the very first month.

  • Corporate accounts with controlled access: goodbye to personal ones.
  • Out with duplicates and licenses nobody uses.
  • Library of prompts and agents by department.

EXAMPLE · WHAT SHOWS UP

EXTRACT · STACK BEFORE / AFTER
Before · 14 overlapping tools1,240 €/mo
After · 6 approved and governed480 €/mo
Direct savings from the first month760 €/mo
STOP 06 · SERVER ROOM · 3–5 WEEKS § GDPR: DATA UNDER CONTROL

Sandbox: the safe for your data

In plain English: we build a digital vault so you can use AI with your sensitive data —customers, finance, health— without it slipping out of control.

  • Private environment in the EU, encrypted and logging every query.
  • Contractual guarantee: your data does not train the models.
  • We unlock the use cases put on hold out of caution.

EXAMPLE · WHAT SHOWS UP

EXTRACT · SANDBOX ARCHITECTURE
Your datacustomers · finance
Private environmentEU servers · encrypted
AI modelsno retention
Servers in the EUYour data doesn't train modelsEvery query logged
STOP 07 · OPERATIONS ROOM · MONTHLY

AI Ops: someone's watching the screens

In plain English: every month someone checks that everything works, fixes what breaks and brings you only the updates that matter.

  • 3-page report: usage, cost and value. Not 30.
  • Maintenance of agents and prompts: they degrade on their own.
  • Office hours for the team's questions, live.

EXAMPLE · WHAT SHOWS UP

EXTRACT · MONTHLY REPORT (JUNE)
Staff actively using AI71% +6
Stack cost468 € −4%
Incidents resolved2 / 2
New updates rolled in3
STOP 08 · LEADERSHIP · MONTHLY

Fractional CAIO: the corner office

In plain English: a part-time AI director who sets the course, chairs the committee and answers to leadership. Without hiring anyone.

  • AI committee: decisions that don't get filed away in a drawer.
  • Living roadmap: we re-prioritize based on results.
  • 1:1 with the general management to align AI and business.

EXAMPLE · WHAT SHOWS UP

EXTRACT · YOUR MONTH WITH AI LEADERSHIP
AI committee · decisions and minutes2 h
Roadmap and prioritization½ day
1:1 with the general management2 h
Asynchronous availabilityIncluded

Cybersecurity, data governance and AI. All three, at once.

There's no useful AI without governed data, and no governed data without security. Every room in the office works on all three fronts.

Cybersecurity

So adopting AI never opens a single door.

  • Shadow AI detected and removed from the stack.
  • Corporate accounts with SSO and access control.
  • Encrypted private environments, fully traceable.
  • A clear protocol for unapproved tools.
STOPS010506

Data governance

Knowing what data leaves, where it goes and under what contract.

  • Map of data flows to each AI provider.
  • DPA signed and reviewed with every tool (GDPR).
  • Sensitive data only in controlled EU environments.
  • The AI systems register the AI Act requires.
STOPS01020406

AI

So artificial intelligence delivers, measured in euros.

  • Consolidated stack and a library of prompts and agents.
  • Use cases prioritized by return, not by hype.
  • Team trained (Article 4) and supported every month.
  • Ongoing strategic direction with committee and roadmap.
STOPS03050708

This isn't optional: what the law already requires.

The AI Act applies in phases, and two of them are already in force. If your team uses AI — and it does —, these obligations are yours even if you've built nothing yourself.

Timeline of obligations for companies that use AI

Up to 35 M€ or 7%of annual global turnover: the AI Act's maximum penalty range.
2 FEB 2025
IN FORCE

Prohibited practices + AI literacy

Social scoring and manipulation are banned. And Article 4 requires training for all staff who use AI. Already enforceable.

2 AUG 2025
IN FORCE

General-purpose models

Transparency and governance obligations for the large models (the ones your tools run on).

2 AUG 2027
FINAL PHASE

High risk in regulated products

AI embedded in products with their own regulations (medical devices, machinery, toys…).

And GDPR still applies to everything: every piece of personal data that goes into an AI tool needs a data processing agreement (DPA) with the provider. A free ChatGPT account holding customer data is a breach today, without waiting for any deadline.

/WE START WHENEVER YOU'RE READY

Shall we open the door? We start at the first stop.

Tell us where you are and we'll tell you which stop to come in at: usually 01 — or straight to 03 if you urgently need to meet the mandatory training requirement. Every stop has a fixed price.

CYBERSECURITY · DATA GOVERNANCE · AI — ACROSS ALL 8 STOPS, AT A FIXED PRICE

\n