Re:amaze

Set up an AI agent for Re:amaze. On a schedule it drafts or sends replies and triages conversations, trained on your templates and past conversations.

eesel's AI agent for Re:amaze works your Re:amaze conversations like a member of your team. On a schedule it drafts replies as internal notes or sends them, sets tags, assignee and status, and learns from your response templates, past conversations and help center articles.

Invite your AI teammate into Re:amaze and it works your conversations like any other agent on your team. It reads your response templates and your past conversations first, so it already knows how your team answers. Then it drafts replies for a person to send, replies itself when you let it, leaves internal notes, and tags and assigns what comes in.

With Re:amaze, an automation runs on a schedule. Your agent checks your conversations as often as every ten minutes and acts on the ones you told it to handle.

In this example a scheduled run found a new customer email and left its drafted reply as an internal note.

A Re:amaze conversation with your agent's drafted reply waiting as an internal note.

Set up eesel in Re:amaze

1

Ask your agent to connect Re:amaze

Open the chat in your dashboard and ask it to connect Re:amaze. Your agent asks for three things: your subdomain, the staff login email the token belongs to, and an API token. Prefer to click? Integrations in the left sidebar has the same form.

Your agent lists the three things Re:amaze needs.

Your subdomain is the first part of your Re:amaze address. Replies and internal notes post under the staff account you connect with, so pick the account you want customers to see. Give the subdomain and the email in the chat. The token goes into a secure form in the next step.

2

Paste the token into the secure form

Once it has your subdomain and login email, your agent opens a secure form for the token. What you type there goes straight to eesel. It never enters the conversation, and your agent never sees it. The form expires after thirty minutes.

The token goes into a secure form, not into the chat.

To get the token, open Settings > Developer > API Token in Re:amaze and copy it. Paste it into the form and click Save securely. It shows as dots, not text.

The pasted token stays hidden, ready to save.

3

Confirm the connection

Saving the token finishes the connection. Your agent confirms which Re:amaze account it connected and which staff login it is working as.

Your agent names the account and the staff login it connected as.

Check it worked. Integrations > Re:amaze in the left sidebar shows the connection with your subdomain. From a terminal, eesel status lists Re:amaze as connected.

4

Check what it can read

Open Integrations > Re:amaze to see what your agent pulled in. It reads two sources: Templates, your response templates, and Conversations, your past conversations. Each source has its own toggle, so you can leave one out without disconnecting. Each shows a document count once its first import finishes. Below the sources sit the actions, grouped as Read and Write.

Each source has its own toggle and its own document count, with the actions below.

Your help center articles are not listed here. Your agent reads them live from Re:amaze while it works.

5

Try it on a real conversation

Before you automate anything, ask your agent to work a conversation in the chat. Ask which conversations have no reply yet, then ask for a draft on one. It reads the thread, checks your templates, past conversations and articles, and shows you the reply it would send.

Ask for a draft and your agent writes the reply it would send.

If no article confirms a detail, it says so instead of guessing. Under the draft it lists the conversations and articles it used.

Every draft lists its sources and says what it could not confirm.

6

Correct it once and it remembers

Tell your agent what to change and it rewrites the draft. Tone, length, sign-off, house rules.

Ask for a change and your agent rewrites the draft.

Then ask it to keep those preferences. It writes them into its own instructions, so every later draft follows them, in chat and on a schedule.

Your agent confirms the rules it saved and links to them.

Check it worked. Your Instructions page holds the rules in plain sentences, and you can edit them there. From a terminal, eesel instructions shows the same text.

The saved rules on your Instructions page.

7

Create your first automation, on a schedule

Connecting alone does not start your agent. It needs an automation first. Ask for what you want handled. For Re:amaze, your agent offers a scheduled automation: it checks your conversations on a timer and acts on the ones that match its instructions. The shortest interval is every ten minutes.

Your agent offers a schedule that checks Re:amaze every ten minutes.

Say yes and it creates the automation: a schedule plus instructions saying which conversations to take and what to do with them. It appears on your Automations page, switched on.

The new schedule on your Automations page, switched on.

Open it to change how often it runs, the timezone, or what it does. On every run your agent checks Re:amaze for conversations that match the instructions and acts on the ones it finds.

The schedule editor, where you set how often it runs and what it does.

Check it worked. Your Automations page lists the schedule, switched on, with its interval under the name. From a terminal, eesel automations lists it as on.

8

Check automation work in Re:amaze

Your Activity page shows each run and what your agent did.

Then wait for the next run and open Re:amaze. In this example a customer emailed the support channel. The next run found the conversation with no staff reply, and your agent left its drafted reply as an internal note.

The customer's message, with your agent's internal note underneath.

The note is internal, so the customer never sees it. A teammate reads it, edits what they want, and sends.

The whole drafted reply, ending in the sign-off your team asked for.

From Claude Code, Cursor or a terminal. eesel login signs you in and eesel init walks the setup. eesel integrations connect reamaze prints the details Re:amaze needs: the subdomain, the staff login email and the API token. A person must copy the token from Re:amaze, and it must not pass through a chat, so do the connect step in the dashboard chat, where your agent opens the secure form and waits for the person. Then eesel status confirms the connection, eesel integrations download start copies your templates and conversations in, and eesel automations shows the schedule once it exists. Full command reference: The eesel CLI. Step-by-step: Set up from Claude Code or the terminal. From a Model Context Protocol client: Claude Code and MCP clients.

Sources, schedules and actions

Sources

Sources are what it learns from:

Source	What your agent reads
Templates	Your response templates, so it reuses a canned reply word for word when one fits
Conversations	Your past conversations, so it learns how your team already answers

Your help center articles are read live from Re:amaze through the List articles action, so a new article is available to your agent the moment you publish it.

Every other integration on your agent is also a source, so a Re:amaze reply can draw on your website, Confluence and your files at once. See Integrations.

Schedules

A Re:amaze automation runs on a schedule. On each run your agent checks Re:amaze for conversations that match its instructions and acts on the ones it finds. You can change how often it runs at any time.

How often	As often as every ten minutes, or hourly, daily, weekly, or a pattern you set
What it handles	Whatever the instructions describe, for example open conversations with no staff reply yet
What customers notice	A new conversation is picked up on the next run, so within ten minutes on the ten minute setting

Anything your agent can do on request it can do on a schedule: drafting notes, replying, tagging, assigning and triage. See Automations and Schedules.

Actions

Actions are what it can do. The integration page groups them as Read and Write. Read actions only look things up. Write actions change a conversation.

	Action	What it does
Respond	Send conversation reply	Sends a reply the customer can see. Before it posts, it reads the thread once more and stops if someone has already answered
	Leave internal note	Adds a private note only your team sees. Start here: your agent drafts, a person sends
Manage	Update conversation	Sets tags, assignee and status
	Create conversation	Opens a new conversation on a customer's behalf
Look up	List conversations	Lists your conversations. It can narrow to open, unassigned or archived ones, or filter by customer email or tag
	Read conversation	Reads a whole thread in order, with each message's author and whether the customer could see it
	List articles	Lists your help center articles
	List response templates	Lists your canned replies
	List channels	Lists your channels, with each one's name and type
	Get contact	Looks up a contact by name or email

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.

Replying does not close the conversation, even though Re:amaze normally does. Tags are only ever added, never removed.

Limitations

• Scheduled, not instant. A new conversation is picked up on the next run, so within ten minutes on the shortest setting.

• Tags are only ever added. Your agent cannot remove a tag.

• Replying does not close the conversation. Ask for the status change separately if you want it resolved.

• Replies post as the connected staff account. To change the name, connect again with that person's token.

• A new token means reconnecting. A rotated or revoked token stops the connection until you connect again with the new one.

• It cannot read a channel's inbound email address. Find it in your Re:amaze channel settings.

Troubleshooting

Nothing happened after a new conversation arrived?

Your agent picks conversations up on the next scheduled run, so on the ten minute setting allow up to ten minutes. Check the automation is switched on on your Automations page, and open your Activity page to see each run and what it decided.

The drafted note never appeared?

Look at the run on your Activity page. It shows whether the note was written, skipped, or is waiting for you to review it.

Your agent says somebody already replied?

Before it posts a reply, your agent reads the thread once more. If the last message the customer can see is already an answer, from your agent or from a person, it stops and tells you. This is what stops a customer being answered twice.

Authorization errors after changing the token?

A rotated or revoked token means connecting again with the new one. Ask your agent to connect Re:amaze and it opens the secure form for you.

Which email address do customers write to?

Your agent cannot read a channel's inbound email address from Re:amaze. Find it in your Re:amaze channel settings. An email to that address becomes a conversation, which the next scheduled run then picks up.

FAQ

How do I add an AI agent or chatbot to Re:amaze?

Ask your agent in the dashboard chat to connect Re:amaze, give it your subdomain and staff login email, and paste an API token from Settings > Developer > API Token into the secure form it opens. Then ask for a schedule.

A good first schedule is "every ten minutes, draft an internal note on any open conversation with no staff reply". Your templates and past conversations import on their own, and your agent works those conversations in Re:amaze beside your team. The steps are in Set up eesel in Re:amaze above.

Does the Re:amaze AI agent send replies on its own, or draft them for review?

Either, depending on how you set it up. Leave internal note writes a private note your team reads, edits and sends, Send conversation reply sends it.

Connecting only gives your agent access to your templates and conversations. It needs an automation before it acts at all.

How quickly does the Re:amaze AI agent pick up a new conversation?

On the next scheduled run, so within ten minutes on the shortest setting. Re:amaze has no event triggers in eesel, so your agent does not reply the moment a message arrives.

You can set the schedule to every ten minutes, hourly, daily, weekly, or a pattern you set.

Can I limit which Re:amaze conversations the AI agent handles?

Yes. Describe the conversations you want in the schedule's instructions, for example "open conversations tagged wholesale with no staff reply", and your agent leaves the rest alone.

Your agent reads each conversation, including who wrote it, before it acts, so you can skip topics or senders in plain words. More options are in Control how your Re:amaze AI agent behaves above.

Can I connect Re:amaze to Claude or ChatGPT?

Yes, through eesel. Once Re:amaze is connected, you can talk to your eesel agent from Claude, ChatGPT or Cursor, and it works with your Re:amaze data.

See Use eesel from Claude and other assistants, Use eesel from ChatGPT and Claude Code and MCP clients. Do the connect step itself in the dashboard chat, so the API token goes into the secure form.

What reporting do I get for the Re:amaze AI agent?

The Activity page lists every scheduled run and what it decided, and the Reports page shows volume, triggers and approvals over time.

Each run shows what it found, each action it took, who approved it, and its reasoning. You can also ask your agent for a written report, or put one on a schedule.

Whose name do the Re:amaze AI agent's replies appear under?

The staff account whose API token you connected. Replies and internal notes post as that person.

To change it, connect again with a token from the account you want.

Can I connect more than one Re:amaze account to eesel?

Yes. Connect each one the same way, and when more than one is connected, your agent asks which account you mean.

Related pages

• Helpdesk for the job your teammate takes on in a ticket queue

• Automations and Schedules for how schedules work, and what else can run on one

• Actions and Approvals for choosing what your agent may do

• Integrations overview for how sources and actions fit together across your apps

• Set up from Claude Code or the terminal