> ## Documentation Index > Fetch the complete documentation index at: https://f4c7a9e2d8b1-docs.tenzo.ai/llms.txt > Use this file to discover all available pages before exploring further. # AI Chat > Ask questions about your candidates, jobs, and the Tenzo platform from anywhere in the app ## What it is AI Chat is an in-product assistant that answers questions about your data and the Tenzo platform. You can open it from any page and ask things like: * *"Find the elementary janitor job — what's its status?"* * *"Pull up Jane Smith's contact info"* * *"Which candidates have been added to the warehouse role this week?"* * *"How do I set up a rule to auto-pause applications?"* The assistant uses Gemini (Google) and is grounded in your organization's live data plus the Tenzo documentation. It only sees data you would normally see in the app — your existing permissions apply. ## How to open it Click the chat icon in the navigation bar to open the assistant in a popup. The popup stays anchored as you move between pages so you can keep the conversation while you work. The conversation lives only in the current session. Closing the tab clears history. ## What it can do The assistant has read-only access to your recruiting data: * **Candidates** — look up by id or by name (case-insensitive substring), see contact info and basic profile fields. * **Jobs** — look up by id or search by name; honors folder-level access. * **Users (your teammates)** — search by first or last name, see email. * **Documentation** — search and read pages in this docs site to answer "how do I…?" questions. When the assistant runs a tool you'll see a small inline status line (e.g. "🔍 Searching candidates for 'janitor'…") so you can tell what it's doing. ## What it can't do (yet) * It can **not** edit your recruiting data. It cannot create candidates, change application stages, send messages, or schedule interviews. All such write actions still go through the existing UI. (It can record its own feedback to an internal log when it's unsure — this never touches your data.) * It does **not** persist history across sessions or across devices. * It can only answer questions about the org you are currently logged in to. ## Privacy and safety * **Org isolation:** The assistant cannot read data outside your organization, ever. * **Job access:** Tool calls that fetch a specific job honor your folder-level access — if your role is restricted to certain folders, the assistant is restricted the same way. * **Candidates and team members:** As in the rest of the app today, these are scoped to your organization but are not separately gated by folder or role — anyone in your org who is allowed to be there can look them up. * **No write access to your data:** The assistant cannot change your recruiting data — any such change has to be done in the regular UI. The only thing it writes is its own feedback to an internal log (questions it struggled with), which never exposes or alters your data. * **Conversation handling:** Conversations are sent to Gemini (Google) for processing. Google does not train on this data. Tenzo does not retain the conversation after the session ends. ## Limitations and tips * Lookups by name are substring matches, capped at 100 results — very generic queries (e.g. "find candidates") will return a slice, not the whole list. Be specific. * The assistant works best for one-off lookups, "how do I" questions, and focused analytics — it can report how many candidates each user sourced over a period, summarize how candidates are scoring on a job's interview questions (including which question is hardest and whether grading looks too harsh or lenient), and surface the top-ranked candidates for a role. For broad, interactive exploration it still isn't a full replacement for the analytics dashboards. * If a question requires data the assistant doesn't have access to (write actions, cross-org, external systems), it will tell you and suggest where to find it in the UI. ## Need help? If the assistant gives you a wrong answer or behaves unexpectedly, please flag it to support so we can improve it.