Went live too soon
Straight from pilot to production, without proper testing. The first error is immediately visible to a customer.
AI & Governance
AI implementation for SMEs
Between “it works in the demo” and “it's part of the day-to-day work” lies the part where things usually go wrong. We help you make that transition, with attention to privacy, security, human oversight, testing, and working agreements people actually follow.
ResponsiblePrivacy, security and compliance from the outset.
People firstHuman review where it matters.
Tested and provenFrom pilot to practice, not from demo to hope.
Implementation overview
example
Pilot completed→
Testing completed→
Working agreements→
Go-live→
Evaluation
Privacy and dataIn orderDPIA completed
SecurityIn orderaccess and logging
Human in the loopActivewith every decision that matters
Adoption and trainingIn progress3 out of 5 teams
Progress
73%
Next step: document working agreements with the team leads.
Responsible AI use
starts with good
agreements.
starts with good
agreements.
What often goes wrong
Six patterns. They rarely cost you the technology, but they do cost you trust.
No rules of the game
It’s unclear who can make decisions, check, and make changes. So everyone does it differently.
Overlook privacy
Personal data or confidential information that hasn't been properly protected. Difficult to fix after the fact.
Output not checked
Answers are blindly accepted without validation. Until one is wrong.
No feedback
User feedback isn't collected, so no one knows where the friction is.
Dependent on one person
Knowledge and administration are in the hands of the enthusiast. If they leave, everything comes to a standstill.
What we focus on
Secure setup
Data sovereignty, privacy by design and security to the standard. See data sovereignty.
Human in the loop
Human review and explanation for decisions that affect people.
Quality control
Testing, validation, and monitoring of output and performance — including after launch.
Clear agreements
Roles, responsibilities, and escalation documented. See AI governance.
Monitoring
Ongoing insight into usage, quality, and impact, rather than a one-off measurement after the fact.
Make adjustments
Regularly evaluate and adjust based on data and feedback.
From pilot to practice
Six steps. Step four is where most engagements get stuck—and where we put in the most time.
1InventoryWe map use cases, data, risks and requirements.
2DesignWe design the solution and define success criteria and metrics.
3Testing and validationWe test with real data, focusing on quality, privacy and bias.
4Working agreementsWe define roles, checks, rules and escalation — this is where things usually get stuck.
5Go-livePhased rollout with training, support, and communication.
6EvaluationWe measure impact, gather feedback and continuously adjust.
The checklist we go through
Every point has an owner. Without an owner it's not an agreement but an intention.
| Control | Status | Owner |
|---|---|---|
| Data classification and DPIA | ready | Privacy |
| Access management and permissions | ready | IT |
| Model selection and documentation | ready | Project management |
| Test plan and validation criteria | in progress | Process owner |
| Check for bias and errors | in progress | Project management |
| Human review in place | in progress | Process owner |
| Logging and audit trail | to be scheduled | IT |
| User training and adoption plan | to be scheduled | Team leadership |
| Incident procedure and escalation | to be scheduled | Executive team |
| Evaluation and metrics | to be scheduled | Process owner |
Weigh risks, don’t avoid them
Likelihood versus impact. Start with the top right; leave the bottom left for later or never.
| Low likelihood | Medium likelihood | High likelihood | Very high likelihood | |
|---|---|---|---|---|
| Very high impact | 2 | 3 | 4 | 4 |
| High impact | 1 | 2 | 3 | 4 |
| Medium impact | 1 | 1 | 2 | 3 |
| Low impact | 1 | 1 | 1 | 2 |
What this delivers
- More confidence — users know it’s set up securely
- Less risk — lower risk of errors, leaks, and compliance headaches
- Faster adoption — clear agreements make things easier, not harder
- Demonstrable — documented and verifiable against internal and legal requirements
Why this isn’t a barrier. Most organizations think governance slows things down. In practice, it speeds things up: when the rules are clear in advance, no one has to ask whether something is allowed each time. That saves more time than it takes.
Ready to implement AI responsibly?
De Frisse Blik shows where you stand with data, AI and governance—and which step makes the most sense to take first.
Free online scanNo obligationsPractical step-by-step plan
Ask this site
Ask your question about AI, the AI Act, integrations or digitalization. The answer comes from the pages of this site, with references to where it can be found.
The way it works now
- The appointment is in one colleague's mailbox
- Data is manually retyped from one system to another
- The person who knows is on vacation
- No one can see afterwards why this decision was made
As it is
- The agreement is stored with the client, findable for everyone
- Once entered, information flows through to the rest
- Work continues, even if someone is absent
- Every decision has a source, an owner, and a date
drag with your finger
What if Peter leaves?
Two out of five rules on this page disappear. That's how it feels when knowledge isn't part of the company.
What is your next step?
Free self-assessmentTwelve click questions, three minutes, immediate score and your biggest risk.
Fresh LookBusiness case, benchmark and priority. Settled once we build after this.
What it costsFixed price per package, no price per user, no hourly billing.
From the fieldWhat changed at companies that started with this.
Also good to know: frequently asked questions · security and data · terms and conditions · partners · research and figures · templates · what's changing · where you start
Where does the process break down in practice?
An AI pilot only adds value if the source data are reliable, someone is accountable and the output is used in the actual workflow.
Direct answer
Start with a concrete business problem and a measurable outcome, not with an AI tool. Arrange data, ownership, risk and human control before you scale up.
What we do
We connect analysis, business intelligence, data and execution so you don't get stuck in advice or loose tooling.
How it works
- We clarify the current situation, data and the real bottleneck.
- We determine the smallest working solution, owner and measurable outcome.
- We implement or prioritize the next step and verify the result.
- We secure what works and stop or fix what doesn't work.
What you get
A concrete analysis, priorities, decision criteria, an executable next step and visibility into what needs to be built, integrated or secured next.
For whom and whom not
Right for Dutch SMB organizations that want to solve a concrete process, knowledge, data or AI problem. Not for organizations looking only for a generic tool demo or unproven AI promise.
Costs and choice
Price and scope depend on the chosen next step. View the current fixed packages and pricing logic at pricing for digitization.
Risks and concerns
We work step by step, make assumptions and ownership explicit, use only the minimum necessary data, and maintain human control over decisions and execution.
Frequently asked questions before you decide
Does everything have to be done at once?
No. The preference is one clearly defined problem with a measurable outcome, then scale up.
What does it cost?
The current pricing and package logic is on the pricing page; custom work is only chosen if the scope really requires it.
How do you prevent a trajectory without results?
By defining the problem, owner, success metric and stop criteria in advance, and doing readback and verification after each step.
How we assess your situation
We map your current process, systems, exceptions and costs. Only then do we advise whether an off-the-shelf solution, a custom integration or another approach makes sense.
From insight to execution
This page supports the following concrete solutions.
Continue in Powerhouse
Choose the next step that fits what you are viewing now.