Assurances: the system, the accounts and your data stay with you.

When choosing a consultant, the first questions are usually about contracts, ownership, privacy and support. I answer them below in two groups: companies and independent experts. These are the principles that also go into the written proposal when we work together.

For companies

Who owns the system, the workflows and the code?

You do. Workflows, prompts, configuration and the code I write are set up in your company's accounts. At handover they stay with you, together with a document that explains how they work.

I avoid vendor lock-in: where possible I choose open-source or widely used tools, so your own team or someone else can maintain the work later.

How do support, maintenance and response times work?

As a boutique consultant who works alone, I do not promise 24/7 on-call support. Instead we agree the following in writing before we start:

• Response time and channel: how quickly I respond to which kind of issue.
• Maintenance scope: monthly monitoring, log reviews, model and tool updates.
• Fallback plan: every agent is built so the work can continue by hand if it stops. Critical steps already go through human approval, so the process does not stop when an agent does.

Will our company data go abroad or to uncontrolled AI servers?

Not without your approval. Depending on how sensitive the data is, we choose together between a business cloud, your own server or a local model running in-house. If a cloud is used, I share in writing where the data is processed and under which contract terms, and choose a route that complies with GDPR and KVKK transfer rules. Details are in the data privacy guide.

Do you sign an NDA?

Yes. Before work starts, or before the first call if you prefer. I also read the NDAs you have with your own customers and build the setup within their limits.

Do we need to replace our ERP, MES or CRM?

No. Agents are built alongside your existing systems. On the shop floor I start with read-only access to the plant network, and I work vendor-neutral.

How do we measure results, and what if the target is not reached?

Before the pilot starts, we agree the baseline, the target and the acceptance criteria in writing. Measurement is weekly. At the end of the pilot the decision is yours: roll out, adjust or stop. If the target is not reached, I report the reasons in writing and we do not roll out.

Who is responsible if an agent makes a mistake?

The setup is built so that the agent never decides on its own. Everything that goes to customers, suppliers or outside the company passes human approval, and steps such as payments, contracts or sending email do not run without it. Every step the agent takes is logged, so any error can be traced to where it happened.

Which standards and frameworks do you use as references?

I draw on the frameworks below when designing and documenting a setup. This is not a certification claim; we assess together whether your work needs any certification.

• AI management and risk: ISO/IEC 42001, ISO/IEC 23894, NIST AI RMF.
• Information security and personal data: ISO/IEC 27001, GDPR, KVKK.
• AI application security: the OWASP top 10 risk list for large language model applications.
• Regulation: the EU AI Act (Regulation (EU) 2024/1689).
• Automotive and quality: IATF 16949, VDA 6.3 (process audit), AIAG-VDA FMEA, APQP, PPAP, MSA, SPC, TISAX (information security), ISO 9001.
• Manufacturing: ISA-95 (system levels), IEC 62443 (industrial cybersecurity), ISO 50001 (energy).

What happens to access and copies when the work ends?

The access I was given is closed and working copies are deleted. You keep a handover document that explains how the system works and how to maintain it.

For independent experts

Who owns my accounts, profiles and content?

You do. Your website, domain, LinkedIn page, Google Business Profile and automation accounts are opened in your name. The texts, images and case studies we produce are yours too.

Do I have to give you my passwords?

No. Wherever possible I work through delegated access: you grant an editor or admin role and remove it when the work is done. I do not keep your passwords.

Will anything be published in my name without my approval?

No. Every text I write is prepared in your voice and published only with your approval. Agents prepare drafts; you press publish.

Do you guarantee first place on Google or being recommended by ChatGPT?

No. Nobody can guarantee the results of search engines or AI assistants, and you should be careful with anyone who does. What I guarantee is measurement: I measure your visibility at the start and report openly on everything I do and what changes afterwards.

Is my clients' and students' information safe?

Personal data is never put into free AI tools. An agent only sees what it needs for its task, and names and contact details are masked where needed. I share in writing which data is processed where, for GDPR and KVKK.

Does the free check store my data?

The site check does not record the address you enter; the result is shown only on your screen. The Visibility Check calculates your answers in your browser and keeps them only in your browser. Unless you choose to send me the result, no information reaches me.

My budget is small. Do I have to sign up for a monthly plan?

No. The check and the assessment are free, and you can stop after a one-off written roadmap. I choose free or low-cost tools. Monthly support is optional.

What happens when we stop working together?

Everything that was set up stays with you and keeps running. I leave a short handover note that explains how it all works, and you remove my access.

I brought these principles from the automotive supply chain.

On Hyundai and Ford programs, written acceptance criteria, confidentiality and traceability were never up for negotiation. I work with the same discipline as a consultant: written scope, a measurable target and every step on record. More about how I work is on the About page.