> For the complete documentation index, see [llms.txt](https://docs.eesel.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.eesel.ai/support/faq.md).
# FAQ
eesel AI FAQ: quick answers on setup, agents, knowledge, human oversight and security, with links to the pages that go deeper on each topic.
eesel AI is an AI teammate that works in your helpdesk, website chat, Slack and inbox, and answers from your own knowledge.
Quick answers to the questions that come up most. For anything about your workspace, your team or your bill, see [Account](/account/account.md). For anything specific to a tool, see that tool's page under [Integrations](/integrations/overview.md).
## Getting started
How do I sign up?
Go to [dashboard.eesel.ai](https://dashboard.eesel.ai/signup) and create an account. No credit card needed. You start with a 7-day free trial with 100 credits and every feature unlocked. See [Pricing](/pricing/overview.md) for what the trial includes.
How long does setup take?
Most teams have an agent answering questions within about 15 minutes. Connect a knowledge source, ask it a few questions in chat, and correct what it gets wrong. Going live takes as long as you want it to, because you decide when to turn on an automation.
Pick the closest [use case](/use-cases/helpdesk.md) and follow its quick start.
Where do I actually start?
Open the chat and tell your teammate what you want. It's the expert on eesel itself, so it can connect your tools, explain what it can do, and set things up with you. Almost everything in these docs can be done by asking.
## Agents
How many agents should I have?
Start with one. A single agent can hold several roles and connect to as many tools as you like, so a support agent that also answers your team in Slack and looks up orders in Shopify is one agent, not three.
Add another when you want a genuinely separate job with its own instructions and knowledge, like a content writer alongside a support agent.
Can my agent handle multiple languages?
Yes. It replies in whatever language the customer writes in. On chat widgets you can also turn on auto-translation so it detects and matches the visitor's language.
Can I control which tickets my agent responds to?
Yes, and this is worth setting up early. An [automation](/automations-and-schedules/automations-and-schedules.md) takes a trigger plus instructions, and those instructions can describe exactly which conversations to handle and which to leave alone. Everything else is skipped, and you can see what was skipped and why.
Will it start replying to customers as soon as I connect my helpdesk?
No. Connecting a tool only gives your agent access. Nothing happens until you turn on an automation, so you can connect everything and test privately first.
## Trust and oversight
How do I keep a human in the loop?
Use internal notes, approvals, or both.
Have your agent write its answer as an **internal note** instead of sending it, so your team reads and sends it themselves. Or require **approval** on specific actions, so the agent prepares what it wants to do and waits for you to approve it once, always allow it, or deny it.
Most teams start with internal notes, then hand over more as they see it get things right. See [Instructions and Memory](/instructions-and-memory/instructions-and-memory.md).
How do I know what it's been doing?
**Activity** in the left sidebar lists every run across every channel, and for each one shows what triggered it, what knowledge it read, what actions it took, and its reasoning. **Reports** shows the pattern over time. You can also just ask your teammate how it's been doing. See [Reports](/reports/reports.md).
What if it gets something wrong?
Tell it, in chat. It turns your correction into a rule saved to its own instructions, so it won't repeat the mistake. That's the fastest way to improve an agent, and it works from day one. See [Instructions and Memory](/instructions-and-memory/instructions-and-memory.md).
## Knowledge
What can it learn from?
Your help center, past tickets, macros, your website, internal wikis like Confluence and Notion, Google Drive, and files you upload. Every integration you connect to an agent becomes another source it can answer from. See [Integrations](/integrations/overview.md).
What file formats can I upload?
PDF, plain text, markdown and CSV. Files need selectable text, so a scanned image inside a PDF won't be read. See [Uploading Files](/integrations/knowledge/uploading-files.md).
Can my agent read a screenshot a customer sends?
Yes, on Zendesk, Intercom, Freshdesk, Freshchat, Gorgias, Slack, Microsoft Teams and Discord. It opens the image, looks at it and answers from what it shows, on every plan with nothing to turn on. Other channels are not covered yet. This is separate from uploading files as knowledge. See [Images and attachments](/integrations/images-and-attachments.md) for the file types and the per-channel detail.
How does knowledge stay up to date?
Helpdesk knowledge updates as your agent works. For other sources you can re-sync from the integration's page, or just ask your agent to re-sync it. If something looks stale, a re-sync is the first thing to try.
Can I train on past tickets?
Yes, with a full helpdesk connection. Turn on the tickets source and you can filter by status and date, so you can focus on solved tickets from a recent period. It's the best way to get your agent sounding like your team.
Can I use eesel from Claude or ChatGPT?
Yes. Your workspace is an MCP server, so you can add it to Claude, Cursor or another MCP client, and eesel has apps for ChatGPT. See [Use eesel from Claude and other AI assistants](/apis-and-developer-resources/use-eesel-from-claude-and-other-assistants.md) and [Use eesel from ChatGPT](/apis-and-developer-resources/use-eesel-from-chatgpt.md).
## Security
Is my data safe?
Your data is encrypted in transit and at rest, never shared between customers, and never used to train AI models. See [Security and Privacy](/account/security-and-privacy.md) for the full picture, including what we can provide for a security review.
Can it be stopped from doing something risky?
Yes. Actions are enabled per agent and each one can be switched off entirely, set to require approval every time, or allowed to run on its own. An action you haven't enabled isn't available to your agent at all.
## Still stuck?
Open the support chat under **Help and support** at the bottom left of your dashboard, or the chat bubble on our website. A real person answers it. See [Contact Us](/support/contact-us.md) for email and sales.
## Related pages
* [Account](/account/account.md) for your workspace, team and billing
* [Security and Privacy](/account/security-and-privacy.md) for how your data is handled
* [Actions and Approvals](/instructions-and-memory/actions-and-approvals.md) for keeping a human in the loop
* [Use eesel from Claude and other AI assistants](/apis-and-developer-resources/use-eesel-from-claude-and-other-assistants.md) and [Use eesel from ChatGPT](/apis-and-developer-resources/use-eesel-from-chatgpt.md)
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.
Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:
```
GET https://docs.eesel.ai/support/faq.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.