> 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/use-cases/helpdesk.md). # Helpdesk Set up an AI agent for customer support in Zendesk, Freshdesk, Intercom, Gorgias or another helpdesk. It drafts replies, routes tickets and uses your knowledge. An AI agent for customer support is an AI teammate that works tickets inside the helpdesk your team already uses. eesel's agent drafts replies for review or sends them itself, triages, tags and routes tickets, and answers from your own help center, macros and past tickets. Your Helpdesk Agent works like a human agent on your team, following the instructions you give it.

eesel AI working as an agent inside a helpdesk, drafting a reply on a ticket — Your agent works inside your helpdesk, reading the ticket and drafting the reply.

**Works with:** [Zendesk](/integrations/helpdesks/zendesk.md), [Freshdesk](/integrations/helpdesks/freshdesk.md), [Salesforce](/integrations/helpdesks/salesforce.md), [Intercom](/integrations/helpdesks/intercom.md), [Gorgias](/integrations/helpdesks/gorgias.md), and the other helpdesks in [Integrations](/integrations/overview.md). ## Set up your AI helpdesk agent {% stepper %} {% step %} ### Talk to it to get started Open the chat in your dashboard and tell it about your company. Give your name, your website URL and your help center. Or ask it to connect your helpdesk. It handles the rest. Each helpdesk offers two ways to connect. Quick start reads your public help center only. A Full connection adds past tickets, macros, triggers and actions. See the integration page for yours.

A dashboard chat where the user asks the agent to connect Zendesk, and the agent lays out the Quick start and Full connection options — Ask in chat and the agent walks you through connecting your helpdesk.

{% endstep %} {% step %} ### Ask it a question Test it with the questions your customers actually ask, or ask it to draft a reply for one of your tickets.

The dashboard chat drafting a reply for Zendesk ticket 3400, showing the commands it ran and the drafted reply about annual school licence billing — Ask for a draft on a real ticket and the agent reads the ticket, checks your knowledge, and writes the reply.

The same chat scrolled down, showing the agent naming the help center article it used and asking whether to post the draft as an internal note or send it as a public reply — It names the help center article it used, then asks whether to post the draft as an internal note or send it as a public reply.

{% endstep %} {% step %} ### Give it feedback Tell it what was wrong, right there in the chat. Too formal, missing a step, wrong sign-off, wrong email address. Here the draft is sent back with four corrections at once.

A dashboard chat where the user asks for no dashes, a Solvy sign-off, the correct support email and a friendlier tone, and the agent returns the rewritten draft — Ask for the changes in plain words and it rewrites the draft, then offers to save the changes as preferences.

Say yes and it writes the corrections into its own instructions, so every later draft follows them.

The agent confirming it has saved the reply style preferences, with a link to the full instructions and follow-up options for ticket 3400 — It confirms what it saved and links you to the full instructions.

Open that link to read the saved rules on your Instructions page.

The Instructions page open beside the chat, showing the saved rule about casual tone, no dashes, the Solvy sign-off and the support email — The corrections written out as a rule on your Instructions page, where you can edit them yourself.

{% endstep %} {% step %} ### Create your first automation Connecting alone does not start your agent. It needs an automation first. Ask it to start drafting replies, or tell it what you want it to handle. It creates an [automation](/automations-and-schedules/automations-and-schedules.md) for you.

The dashboard chat where the user asks for internal note drafts on every new ticket, and the agent reports the trigger and action it enabled — Ask in chat for the automation you want, and the agent builds it.

Open the automation to see or change what it does. The trigger event is fixed when the automation is created. So a new automation is required for each different trigger event.

The automation page in eesel showing its name, the Zendesk event it runs on, and the instructions telling it to draft a reply as an internal note — The automation's own page: its name, when it runs, and the instructions it follows.

Then check the automation's work in your helpdesk. Here the next Zendesk ticket in has the drafted reply waiting as an internal note, signed by eesel AI, for a teammate to check and send.

A Zendesk ticket where eesel AI has posted its drafted reply as an internal note, with troubleshooting steps and a line saying it was left by eesel AI — The drafted reply waiting as an internal note on the next Zendesk ticket.

{% endstep %} {% endstepper %} {% hint style="info" %} **From Claude Code, Cursor or a terminal.** The same setup works as commands: `eesel login`, then `eesel integrations connect gorgias` (a person approves it in the browser; if the receipt has no link, ask `eesel chat` for one), `eesel integrations download start` to copy your sources in, and `eesel automations` to see and switch on the automations. Swap `gorgias` for `zendesk`, `freshdesk` or `intercom`. Trigger keys are in the Triggers table on each integration page. Full command reference: [The eesel CLI](/apis-and-developer-resources/cli.md). Step-by-step: [Set up from Claude Code or the terminal](/getting-started/set-up-from-a-coding-agent.md). From an MCP client: [Claude Code and MCP clients](/apis-and-developer-resources/claude-code-and-mcp-clients.md). {% endhint %} *** ## What it can do * **AI draft replies**, as internal notes for review, or sent straight to the customer * **Tag, categorize, and update tickets**, including fields and status * **AI ticket routing** to your team with the assign actions, driven by your own instructions, for example low confidence or an angry customer * **Answer in the customer's language** without translating your help center * **Match your tone and voice**, described in plain text

## Your team stays in control Nothing runs until you create an automation. You stay in control the whole way. Adjust your AI agent's access to knowledge and its action permissions to your liking. See [Actions and Approvals](/instructions-and-memory/actions-and-approvals.md). Most teams turn it on in stages: 1. **Start with drafts.** Every reply is a draft or an internal note for a person to check and send. 2. **Let it answer the easy ones.** Once it gets the routine tickets right, let it handle those end to end. 3. **Turn on more when you are ready.** You decide when, and you can switch anything back off at any time.

The dashboard showing an approval request, Allow eesel to use Leave public reply, with the drafted reply and Approve, Always Allow, and Deny buttons — When an action needs your OK, it waits for your answer: approve once, always allow, or deny.

## Correcting the agent * **Give feedback in chat.** It updates its own instructions or adds the missing knowledge. See [Instructions and Memory](/instructions-and-memory/instructions-and-memory.md) * **Edit the instructions yourself** on the Instructions page * **Give it the docs you give your team.** Upload the procedures document you would hand a new hire * **Run a** [**Simulation**](/skills/skills.md)**.** It replays your past tickets and scores its answers against what your team actually sent ## Browser extension [eesel: AI agents for customer service & more](https://chromewebstore.google.com/detail/eesel-ai-chatgpt-sidebar/ejhkkbilnpifailgngpkgmiofhioacjd) is the browser extension. It puts the same agent in a Chrome side panel, beside whatever page you have open. Open a ticket, ask for a draft, and it writes one from the ticket and your knowledge. Then paste it into the reply field, edit it first, or ask for a rework. On a Zendesk ticket the panel offers a **Paste into Zendesk** link that drops the draft straight into the reply field.

A Zendesk ticket with the eesel side panel open, showing the reply it drafted and a Paste into Zendesk link — The side panel drafts the reply from the ticket you have open, and **Paste into Zendesk** drops it into the reply field.

See [Browser extension](/use-cases/browser-extension.md) for everything it can do. ## Skills [Skills](/skills/skills.md) are extra workflows on top of the day-to-day ticket work. Run them on demand, on a schedule, or on a trigger. Ask the agent in chat which skills to run, or ask it to create one. The ones most helpdesk teams start with: * **Simulation**: replays past tickets and scores your agent's answers * **Analyze ticket trends**: recurring topics, sentiment, and volume * **Review agent setup**: audits your configuration and suggests improvements Ask for a skill in chat by name. Here the Helpdesk simulation skill is asked for, and it pulls your resolved tickets and works through them in batches. This run sampled 20 resolved tickets.

A dashboard chat where the user asks to run the Helpdesk simulation skill, and the agent reports the commands it ran while sampling 20 resolved tickets — Ask for the skill by name and it starts work in the chat, telling you each step as it goes.

It answers in the chat: how many of its replies matched what your team actually sent, and how it did per topic. In this run 17 of the 20 matched.

The chat showing the simulation result, 17 of 20 tickets matched the team's reply quality, with a score for each ticket theme — The score for the run, then the same score broken down by ticket topic.

Scroll down for the fixes it suggests, ranked, and a link to the full report.

The same chat scrolled down, listing five ranked fixes and linking the full simulation report file — The ranked fixes it suggests, and the link to the full report.

Open the report to read the run ticket by ticket: what the customer asked, what your team replied, and how the agent's answer compared.

The simulation report open in the dashboard, with match counts, results by theme, and one ticket shown with the customer message and the team's reply — The full report, with the counts up top and every ticket written out below.

## FAQ

How do I add an AI agent to my helpdesk?

Open the chat in your eesel dashboard and ask it to connect your helpdesk, then approve the connection in the browser. After that, ask it to create an automation, for example "draft a reply to every new ticket", and it starts on the next ticket. The steps are in [Set up your AI helpdesk agent](#set-up-your-ai-helpdesk-agent).

Which helpdesks does the AI agent work with?

eesel works with Zendesk, Freshdesk, Intercom, Gorgias, HubSpot, Help Scout, Front, Salesforce, Jira Service Management, Zoho Desk and Re:amaze. Each has its own page in [Helpdesks](/integrations/helpdesks.md) with what the agent can read and do there.

Does connecting my helpdesk start replying to customers?

No. Connecting only lets your agent read your helpdesk. It does nothing on tickets until you create an automation.

Can the AI agent send replies on its own, or only draft them?

Both. It can leave a draft as an internal note for a person to check and send, or send the reply to the customer itself. Most teams start with drafts and let it answer the routine tickets itself once it gets them right. See [Your team stays in control](#your-team-stays-in-control).

How much does an AI helpdesk agent cost?

eesel charges per ticket it handles, once per ticket however many replies it takes. There is no seat fee and no monthly minimum, and every workspace starts with $50 of free usage. See [Pricing](/pricing/overview.md) for the price per task.

Does it learn from our past tickets?

Yes, on a Full connection. It reads your help center, macros and past tickets so it knows how your team answers before it touches anything. A Quick start connection reads your public help center only.

Can it answer customers in other languages?

Yes. It answers in the customer's language without you translating your help center.

Can I use my helpdesk agent from Claude or ChatGPT?

Yes. You can add your eesel workspace to Claude or ChatGPT and talk to the same agent from there. See [Use eesel from Claude and other AI assistants](/apis-and-developer-resources/use-eesel-from-claude-and-other-assistants.md), [Use eesel from ChatGPT](/apis-and-developer-resources/use-eesel-from-chatgpt.md), and [Claude Code and MCP clients](/apis-and-developer-resources/claude-code-and-mcp-clients.md).

## Related pages * [Integrations overview](/integrations/overview.md) for every helpdesk your teammate can join * [Automations and Schedules](/automations-and-schedules/automations-and-schedules.md) for triggers, instructions and schedules * [Actions and Approvals](/instructions-and-memory/actions-and-approvals.md) for choosing what your agent may do * [Instructions and Memory](/instructions-and-memory/instructions-and-memory.md) for correcting your teammate on a live ticket * [Set up from Claude Code or the terminal](/getting-started/set-up-from-a-coding-agent.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 dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.eesel.ai/use-cases/helpdesk.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. 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.