Your clients' AI questionnaire, filled in from what is read.
A large account asks you which AIs process its data, where they send it, who authorized them. Answer from your tools, not from memory.
The question
“A client sends us an AI questionnaire, and nobody really knows what to answer.” Large accounts add an AI section to their supplier questionnaires: which tools, which data, which subprocessors, which internal rules. The answer is often reconstructed from memory, department by department, and it goes stale as soon as a vendor turns on a new feature.
With Ojja Watch
- A ready file, drawn from the register: the AI tools in place, what they read, where the data goes, whether they train their models on it, each vendor's AI subprocessors.
- Each answer carries its source and its date: read in your tools, or declared by you.
- Your internal rules, linked to the tools concerned, when you have written them.
- If you publish software, a section on the AI in your own product, which you fill in.
- A file that follows your tools: when a vendor turns on a feature, the next version of the file says so.
The proof
The free readout, read in your tools: the file is built on the same reading.
What Ojja does not do
Ojja does not answer for you and does not say whether your answers satisfy your client or a text. It gives you the facts, dated and sourced; you sign your answers.
Frequently asked questions
What does a client AI questionnaire ask?
Most often: the AI tools used, the data they process, hosting and transfers outside the European Union, training of models on the data, AI subprocessors and internal rules.
How do you answer without reconstructing everything by hand?
By starting from what your tools already record, read and dated, rather than from a survey of every department.
What if we publish software with AI ourselves?
A dedicated section describes your product's AI, from what you declare.