> ## Documentation Index > Fetch the complete documentation index at: https://docs.helpalive.com/llms.txt > Use this file to discover all available pages before exploring further. # HelpAlive FAQ: the AI agent, chat, assists, users and data > Short answers about the HelpAlive agent in your web app: screenshots, the script, who sees the chat, assists, limits, users' data and speed. This is the HelpAlive FAQ: short answers to the questions teams ask most about HelpAlive, each with a link to the full page. It is for anyone evaluating or setting up HelpAlive. ## What is HelpAlive? HelpAlive is an AI agent inside your web app. It answers your users' questions from your docs, and does tasks for them or shows them how, one step at a time. You add one script and one `identify()` call. See [How HelpAlive works](/product/how-it-works). ## Does it take screenshots? No. The agent reads the page as structured text: headings, visible text, and each button, link and field with its label, state and value. It never takes screenshots or records the screen. See [Data we collect](/trust/data). ## Does it work with any framework? Yes. It works on any web app in the browser: React, Vue, Angular, Svelte, Next.js, server-rendered apps and site builders. In single-page apps it follows page changes on its own. See [Install the script](/installation), or the step-by-step pages for [React](/sdk/react), [Next.js](/sdk/nextjs), [Vue](/sdk/vue) and [Angular](/sdk/angular). ## Do I need to tag elements? No. There's nothing to tag and no screens to describe. Plain, visible labels on your buttons and fields help the agent, as they help screen readers. See [Make your app easy for the agent](/sdk/agent-ready). ## Can it make changes for users? Yes, if you allow it. **Does the steps** on **Controls** means that it clicks, types and saves for the user, and hands back what only they can do. **Shows how** means that it highlights each control and the user does every click. See [Assists](/product/assists). ## Will it delete things without asking? It can press Delete when the task asks for it. When a user asks to delete records by a rule, such as "delete all drafts", it lists them and asks once first. See [Safety and hand-backs](/product/what-it-wont-do). ## Will it pop up on its own? Only when a user gets stuck, such as when a save fails or an error appears. A short note offers help, and clicking it opens the chat with their question already written. Ignoring it does nothing. You can turn nudges off on **Controls**. See [Nudges](/product/nudges). ## Can I limit who sees it? Yes. On **Controls**, choose **Everyone**, selected tenants or people, or a percentage of them. You can also set limits per person, per tenant and for your organization. See [Controls](/dashboard/controls). ## Does it work for visitors who aren't signed in? No. The agent appears only for people your app identifies with `identify()`. See [Identify users](/sdk/identify). ## Where do the answers come from? From the websites and files you publish on **Train**. Where your docs are silent, the agent can use general knowledge of your product, and your docs win when the two disagree. Drafts never reach your users. See [Chat](/product/chat). ## What happens at my limit? Your users never see a limit named. **How far over your plan** on **Controls** sets how far past your plan usage may go. See [Limits and usage](/dashboard/usage). ## Is my data used to train models? The model provider that writes answers and runs assists doesn't train on data sent through its API. We may use the kept words of chats and assists, de-identified, to improve HelpAlive for all customers; write to [support@helpalive.com](mailto:support@helpalive.com) to opt out. See [How the AI uses your data](/trust/ai-and-your-data). ## How do I remove a user's data? Go to **Settings → Privacy Management**, find the person or tenant, and press **Delete their data**. Their usage data and their chat and assist text are deleted; the counts stay, made anonymous. **Stop tracking** means that their new events are dropped instead. See [Privacy Management](/dashboard/privacy-management). ## Can my team read users' chats? Yes. Everyone on your team can read chat transcripts and assists on **Audience**. Tell your users so in your own privacy notice. See [Audience and Reports](/dashboard/audience-and-reports). ## Does it slow my app? It is built not to. The script doesn't hold up your page, because it loads with `async`. The chat widget loads only after `identify()`, and the code for assists loads only when one starts. See [Install the script](/installation). ## Does it use cookies? No. The script sets no cookies. It keeps ids and settings in your site's browser storage, and `reset()` at logout clears the chat entries. See [Data we collect](/trust/data). ## Next From sign-up to the agent answering. Tenant, assist, plan and the rest. Keys, verified users and sign-in. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.