A little room for doubt

Oren is an agentic AI. It can answer questions, use connected tools and carry out tasks, but it can also make mistakes.

Why this happens

AI generates responses from patterns, the context you provide and information it retrieves. It does not automatically know whether every statement is true. Missing context, ambiguous requests, outdated sources and tool errors can lead to confident answers with incorrect facts, invented references or flawed reasoning. This is often called a hallucination.

How to reduce mistakes

• Give clear instructions and relevant files. Explain what a correct result needs to include.
• Ask for sources and open them yourself. Check names, dates, calculations and quoted passages.
• Review drafts and proposed actions before relying on them or allowing external changes.
• Use qualified professional advice for medical, legal, financial or other consequential decisions.
• If something looks wrong, correct the context and ask oren to check it again.

Before you act

Incorrect answers or actions can cost time or money, expose information, or affect people and projects. Review outputs according to the stakes; do not treat AI responses as guarantees. You remain responsible for deciding whether to use an output and for checking important actions. No amount of careful prompting removes every risk.

Found a problem? oren@prepx.space