Orkeste agents read each new Zendesk ticket or Intercom conversation, search your help pages in Confluence or your synced documents, and draft an answer or an internal note. Urgent cases are flagged to the team in Slack, and the replies you choose wait for a person on your team to approve them.
Password resets, invoices, how to export data: half the queue is questions your documentation already answers, and each one still takes a person a few minutes.
An outage report or an unhappy key account looks like any other ticket until someone opens it, sometimes hours later.
The answer to the tricky question lives in a Confluence page nobody finds, or in the memory of the one person who handled it last time.
They read, search the documentation and draft. The replies you choose wait for a person on your team to approve them.
Reads each new ticket, identifies the topic and the urgency, and adds an internal note in Zendesk with a suggested category and priority for whoever picks it up.
Searches your documentation and writes a reply grounded in it. The reply is posted to the ticket or the Intercom conversation once a person approves it.
Spots tickets that mention an outage, a refund or a cancellation and posts them to the support channel in Slack with a one-line summary.
Summarises long Intercom conversations for the next person who picks them up, or for the weekly review in Notion.
The answer drafter on a common question.
Refunds, exceptions, upset customers and anything the documentation does not cover. The agents prepare and flag; your team owns the conversation.
No. They read, search and draft so your team spends its time on the cases that need judgement. The replies you choose wait for a person to approve them.
Not today. In Zendesk an agent adds public replies and internal notes; it does not assign, merge, tag or close tickets. In Intercom it replies to conversations but does not assign, tag or close them.
In the knowledge you give it: Confluence pages, documents synced from SharePoint or OneDrive, and files you add to the agent. Its briefing can tell it to answer only from those sources and to escalate when they do not cover the question.
You edit it or reject it in Validations before anything reaches the customer.
Each run uses credits from your plan, from 2 credits with Claude Haiku 5.5 to 235 with Claude Fable 5.1 depending on the model you choose, and you can set a daily and a monthly budget per agent.