Customer support

A customer service knowledge base your agents can ask

Reps ask in Slack and get the answer from your help center and policies, with the page it came from. When a page is out of date, one sentence fixes it.

Free trial on install. No card needed.

Sample data

One job, in a thread

Day 40, and the customer wants a refund

Ask what to tell the customer. The agent reads the conversation and the policy and drafts an internal note. Sending is a separate step behind an Approve card.

  • Draft only: nothing is sent to the customer.
  • Source shown on every answer.

Sample data

Read the ticket, send the reply

Ask about a ticket and the agent reads it from Zendesk. A public reply shows the exact sentence on an Approve card first.

Sample data

What it does for this team

Answer from the page
Policies and how-tos with the page they came from.
Fix the page
Tell the agent what changed; it updates the page after an Approve.
Read the ticket
Status and details from Zendesk or Intercom in the thread.
Draft the reply
An internal note first. Anything a customer can see waits for an Approve.
Chat widget
Your help center answers visitors first and hands over to a person.

Questions

How does an AI agent help a customer support team?

It answers a rep's question from your help center and policies with the source shown, drafts the internal note, and sends a customer reply only after an Approve.

Can it read our Zendesk or Intercom?

Yes. It can read tickets and conversations, add an internal note, and reply. A public reply always waits for an Approve card.

Does it answer customers directly?

The website chat widget answers visitors from your public help center only, with the source shown, and hands off to a person when asked.

How do we keep the knowledge base current?

When a page is out of date, say so in the thread and ask the agent to fix it. It shows the change on an Approve card and the edit is under your name.

Put the agent on your team

Add it to Slack and ask it one real question.