Two kinds of use
Company agents on one side, the AI tools employees pick up on their own on the other.
One inventory of every AI in use, clear owners, controls set by risk and a go/no-go path from pilot to production. We design it, build it into the platform and run it with you.
Most companies do not lack a policy. They lack someone to turn it into a platform and a process that people actually follow.
Company agents on one side, the AI tools employees pick up on their own on the other.
Agents act in real systems, and unapproved tools spread quietly across teams.
The EU AI Act, KVKK and ISO/IEC 42001 ask for records, risk checks and oversight.
Without a clear go/no-go, good pilots never reach production.
Owner, value and risk differ, so the rules do too. A personal tool that proves its value moves to the company line.
Agents the company builds and runs, such as a support or forecasting agent. Focus: inventory, return on investment, go/no-go and monitoring.
Chat assistants and other tools people use in their daily work. Focus: what they may be used for, which data may go in, and training.
The more an AI does on its own, the more testing, oversight and monitoring it gets.
A person asks, the AI helps. The person uses the answer.
A person starts a task, the agent carries it out.
The agent works alone and waits for approval at critical steps.
The agent runs without a person in the loop. People watch and can stop it.
Autonomy and impact set the level: customer contact, personal data, financial decisions and brand reputation. Each level includes the one before it.
An approved tool list, acceptable use rules and data classification. For every use of AI.
An inventory entry with a named owner, a risk and impact review, and testing before release.
Human approval or oversight, continuous monitoring with alerts, incident handling and an audit trail.
Built on one AI inventory and aligned with ISO/IEC 42001, the NIST AI RMF, the EU AI Act and KVKK.
One front door for new ideas, a return and risk score for each, and a go/no-go decision.
Roles and responsibilities, an AI board or centre of excellence, and a network of champions in every team.
Responsible AI principles, risk and impact reviews, and human oversight where it matters.
A reference architecture, separate environments for building and running, and a catalogue of reusable parts.
Monitoring and alerts, quality assurance and a plan, do, check, act cycle.
Lifecycle, policies and procedures, records and audit, holding the five pillars together.
Not a ladder to climb, a deliberate choice. The goal is not to make every team AI-first, but to put each one on the right model.
Current ways of working stay as they are.
AI supports people in the work they already do.
People and AI work side by side.
Processes are designed around what AI can do.
A maturity score and a roadmap.
The operating model and the policy set.
Environments and monitoring, proven on a few real use cases.
A champion network and training across teams.
Pilots do not stay on paper: each one goes live through the new process.
We do not write a policy and leave. The rules live in the platform your agents run on: who may see what, who approves which step and what every agent did.
Every governance project starts with a maturity score and a roadmap.
AI Check UpAgents we build together pass the same risk review and go/no-go before launch.
AI CraftingMonitoring, audit trail and tested versions keep live agents inside the rules.
Continuous TrainingA baseline, yes: an approved tool list, data rules and an owner. The rest grows with the number and autonomy of your agents.
Yes. Personal tools get their own rules on use, data and training, and the ones that prove their value move to the company line.
No. We set up the inventory, environments, monitoring and approval steps in the platform, and run them with your team.
The model is aligned with ISO/IEC 42001, the NIST AI RMF, the EU AI Act and KVKK, and adapted to how your company works.
With a Check Up. It scores your maturity and gives you a roadmap. You can then start with one step and widen the scope together.