Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.remark.ai/llms.txt

Use this file to discover all available pages before exploring further.

Remark bills based on engagements — meaningful, customer-facing conversations where Remark’s AI actively participates. This applies uniformly across every channel: live chat, email, SMS, and social media. Activity that doesn’t meet the engagement threshold is classified as an interaction, which is unlimited and free on every plan.

Engagements

Billable conversations where Remark AI responds to a customer. Counted toward your plan’s usage.

Interactions

All other activity — routing, tagging, short clicks. Currently unlimited and free on every plan.

What Counts as an Engagement

An engagement is a customer-facing conversation with at least one customer message and one Remark AI response.
SourceEngagement criteria
Chat widgetAt least one reply from Remark AI
Inline activatorAt least 3 total messages in the conversation, with at least one from Remark AI
One engagement per conversation. A single chat session or ticket counts as at most one engagement, regardless of how many messages are exchanged within it.

Why the inline activator has a 3-message minimum

When a customer clicks an inline activator question (e.g., “What sizes are available?”), the system generates an automatic prompt and response. We don’t count that as an engagement unless the customer continues the conversation — at least 3 messages total signals genuine engagement rather than a casual click.

What Counts as an Interaction

An interaction is any Remark activity that does not include a customer-facing AI reply. Interactions are currently unlimited on every plan, though this may change in the future. Common examples:
  • A customer clicks an inline activator but doesn’t continue the conversation (fewer than 3 messages)
  • A chat or ticket where only your human agents respond — no Remark AI involvement
  • Remark takes behind-the-scenes action on a ticket (applying tags, routing, adding internal notes) but does not send a reply to the customer
  • Automated ticket classification or prioritization without a customer-facing message
Think of it this way: if the customer never sees a response from Remark AI, it’s an interaction — not an engagement.

Example Scenarios

Chat widget — full conversation

A customer opens the chat widget and asks about return policies. Remark AI replies with the return window and instructions. The customer asks a follow-up, and Remark responds again.Result: 1 engagementThe chat had at least one customer message and one Remark reply. The entire session counts as a single engagement regardless of how many follow-up messages are exchanged.
A customer clicks an inline activator question (“Is this item in stock?”) on a product page. Remark generates an automatic response. The customer reads it and leaves.Result: 1 interaction (not billed)The conversation has only 2 messages (the auto-generated prompt and Remark’s reply), which is below the 3-message threshold for inline activator chats.
A customer clicks an inline activator question, gets a response, then asks “Do you have this in a size 10?” Remark responds with availability details.Result: 1 engagementThe conversation now has 4 messages total (above the 3-message minimum) and includes a Remark AI reply — this qualifies as a genuine engagement.
A customer emails asking about order status. Remark AI sends a customer-facing reply with tracking information.Result: 1 engagementThe email thread includes at least one Remark AI reply to the customer.
A support ticket comes in. Remark reads the content, applies a “Billing” tag, routes it to the finance team, and adds an internal note — but doesn’t send a reply to the customer.Result: 1 interaction (not billed)Remark took action behind the scenes, but the customer never received a response from Remark AI.
A customer starts a chat. A human agent on your team handles the entire conversation. Remark AI is not involved in the reply.Result: 1 interaction (not billed)Since Remark AI didn’t send a customer-facing response, this is an interaction.
A customer texts your support number asking about store hours. Remark AI responds with hours and the store address.Result: 1 engagementSame rules apply across channels — a Remark AI reply to the customer counts as an engagement.
A customer sends a DM on Instagram asking if an item ships internationally. Remark AI replies with shipping details.Result: 1 engagementSocial media channels follow the same engagement criteria as all other channels.

Quick Reference

ActivityBilled?Why
Chat widget with Remark AI replyYes (1 engagement)Customer-facing AI response
Inline activator click, no follow-upNo (interaction)Under 3-message threshold
Inline activator with continued conversationYes (1 engagement)3+ messages with AI reply
Email with Remark AI replyYes (1 engagement)Customer-facing AI response
Ticket routed/tagged but no reply sentNo (interaction)No customer-facing AI response
Chat handled entirely by human agentsNo (interaction)No Remark AI involvement
SMS with Remark AI replyYes (1 engagement)Customer-facing AI response