Feature

Define what is worth joining.

Conversation labels are tests you define for each stand-in: high-intent buyer, confused evaluator, competitor comparison, painful workflow, or language your team should learn from. Turn on notifications for the labels a rep may want to join live.

What it does

Labels tell Stand what your team considers signal. Instead of asking reps to watch every AI-handled chat, each agent can carry a small set of labels such as High-intent buyer, Confused evaluator, Design partner, or Competitor comparison.

Each label has a name, a recognition prompt, an active switch, and an optional notification switch. When the stand-in recognizes the situation in a real conversation, it privately applies the label, saves a summary, and can notify the agent owner.

The notification does not force a handoff. It brings the rep to the moment that matters. The rep can review the context, choose Offer to join, and Stand asks the visitor naturally before handing the thread over.

When it helps

  • You want founder-quality learning without watching chat all day. Label the conversations a founder, PM, or engineer should hear directly, then let Stand nudge them only when those moments happen.

  • Your product story is still changing. Create labels for confusion, repeated objections, competitor comparisons, or unexpected use cases so the raw visitor language is easy to find.

  • Different reps care about different moments. A sales rep can watch for buying intent, a PM can watch for workflow pain, and a founder can watch for design partners.

  • You want consent-based human entry. A notification can lead to an offer to join, but the visitor meets the rep only after the stand-in asks and the visitor agrees.

  • You need reviewable evidence later. Label matches remain visible in active chats, history, and AI digest summaries with a link back to the conversation.

How to use it

  1. 01

    Open the agent skills

    Go to Agents, open an agent, and switch to the Skills tab. Labeling is available on Pro and Business.

  2. 02

    Enable Labeling

    Turn on the Labeling skill. Pro agents can have up to 3 active labels; Business agents can have up to 10.

  3. 03

    Write the label

    Give the label a short name and tell the stand-in how to recognize it.

    Example: Confused evaluator- “Apply this when the visitor is unsure what Stand replaces, compares it with a support tool, or asks basic questions that show the positioning is not clear.”

  4. 04

    Choose whether it should notify

    Keep some labels as log-only research markers, and turn on Notify on first matched label for the moments a rep may want to join live.

  5. 05

    Let the stand-in apply labels

    During AI-handled conversations, the stand-in privately applies matching labels and saves a short summary with the transcript.

  6. 06

    Offer to join when it matters

    When a notification brings you into a chat, review the context and choose Offer to join. Stand tells the stand-in you are available, weaves that into the current topic, and asks the visitor if they want to meet you.

  7. 07

    Review matches later

    Use History and AI digest summaries to scan label matches, read the original transcript, and turn repeated language into better product, positioning, and follow-up.

Questions

Who decides which conversations should notify a rep?

Your team decides. Reps define the custom labels on their agents, write the recognition prompt, and choose which labels should send a notification when the stand-in matches them.

Does a label automatically put a human into the chat?

No. A notified rep can open the chat and offer to join. Stand then asks the visitor in context, and the handoff happens only if the visitor agrees.

Where do matched labels show up?

Matched labels show on active AI session cards, in conversation history, and in AI digest summaries when AI analytics are enabled.

Try it on your site. Free, forever.