Education & Enablement
AI Training
We train teams to use generative AI on their actual work, knowing what data may go in and how to check what comes out.
The problem
Our people already use AI tools, and we have no idea what data goes in or who checks what comes out.
AI use usually starts off the books: a contract summarized on a personal account, a customer list pasted in for cleanup. Banning tools rarely stops this. It just hides it.
Two risks matter most. Personal data or trade secrets can enter a tool whose terms nobody has read, raising questions under KVKK, Türkiye's personal data law, such as cross-border transfer. And these tools can invent a source, a statute, or a figure and state it with total confidence.
Prompt tricks address neither. The useful questions are who may enter what, and who checks the output before it leaves the building.
You may recognize this
- Staff process work content through personal accounts.
- No written rule says which data may go into which tool.
- AI-drafted text reaches clients with no visible review.
- Some teams lean on these tools heavily while others avoid them.
How we work
Usage and data inventory
By role, we map which tasks already involve AI and what data they touch, flagging anything with customer, patient, student, or employee data.
Data classes and ground rules
With your security and legal people, we agree which data class may enter which tool and fit it on one page. We check account type, retention, and model training settings.
Practice on de-identified work
Participants use their own tasks with identifying details removed, learning to draft faster and to spot where the tool goes wrong.
Verification habits
We produce hallucinations live, then drill checks on sources, figures, quotes, and dates. Every output leaving the company keeps a named human owner.
Spaced review
Weeks later, teams bring the examples and errors they've collected, and we update the rules and prompt library together.
What we deliver
- Role-based usage and data inventory
- Draft AI use policy for your legal team's approval
- Data classification table
- Output verification checklist
- Prompt and workflow library by role
How we measure
- Use of unapproved tools for work data, tracked through anonymous surveys so people answer honestly.
- Verification in practice, checked by sampling outputs for source and figure review.
- Use cases that become routine, which shows experimenting turned into working differently.
Patient letters drafted on free accounts
Illustrative scenario: Administrative staff at a private clinic group use free chat tools to draft patient letters, and the inventory shows treatment details going into personal accounts. The clinic approves one setup and a de-identification rule. In training, staff practice writing requests without patient data and checking every procedural claim against internal sources.
Common questions
Will you tell us which tool to buy?
We help you set selection criteria from your use cases, data classes, and budget. Your security team should review terms and settings, and the final call is yours.
Does this training make us KVKK compliant?
No. Training builds awareness and habits. Compliance depends on your data inventories, contracts, and legal review, and nothing we teach counts as legal advice.
How often do these tools get things wrong?
It varies by tool and task, and no reliable single number exists. Wrong answers read as smoothly as right ones, so sources, figures, and regulations always get checked.
Before we work together, let's think together.
Requests are answered by the founder.