Post-call instructions

How it works

1. 01

   The call is over for good

   Once per call, when no more attempt is planned: never between two attempts. Details just below.

2. 02

   The agent reads the call

   The transcript, the outcome, the summary, the collected information, the call's variables and what was already done during the call (an SMS sent, a meeting booked, a transfer).

3. 03

   It applies your instructions

   For each instruction whose condition is met, it uses the action you named. The others are skipped.

4. 04

   You see what it did

   Every action shows up in the call details, successful or not, with its recipient.

When the instructions run

A call can take several attempts (voicemail, no answer, callback requested). The post-call instructions run only once, on the last attempt, when the call is over.

Someone picks up and the conversation happens

Right after that attempt.

Voicemail or no answer, with more attempts planned

Nothing yet: the call is tried again as set in the Scheduling tab.

Voicemail or no answer on the last planned attempt

After that last attempt, once. That is where an "I could not reach you" text belongs.

The prospect asks to be called back later

Nothing before the callback: they run after the callback call. If no attempt was left, right after this call.

The prospect hangs up

Right after: the call is not tried again.

Inbound call

After every call received.

Someone on your team took over the call, or it was transferred to a human

Never: whoever ran the call already knows how it went. Only the webhook is sent.

Writing good instructions

Variables

Written between double braces, they are replaced by their value before the agent reads your instructions. Three sources, all available:

{{first_name}}, {{company}}, {{city}}…

The call's variables (prompt variables): the ones sent with the call, by the campaign file or the API. The same as in the system prompt, under the same names.

If the prospect is interested, notify {{sales_rep_email}} with @notify_team_by_email, reminding them that {{first_name}} is from {{company}}.

{{outcome}}

The call outcome, as the dashboard shows it (for example "Qualified").

{{outcome_summary}}

The call summary.

{{field_name}}

Each collected field ("Collect information" tab), by its name: {{budget}}, {{callback_date}}…

The agent also sees all these values without braces, along with the contact's name and email: a condition can read "if the city is Paris" without {{city}}. When a collected field has the same name as a call variable, the collected value wins.

Available actions

Everything the agent can do after the call. Type @ in the editor to insert them.

Which email to use

There are two email actions, and the choice depends on who reads the email. The internal report and the link to the call never reach the prospect.

Complete examples

Whole instructions, to copy into the Post-call instructions tab and adapt.

FAQ

The agent did nothing after my call. Why?

No condition was met, or the action was not named with @. Open the call details: they show what the agent did, or why it did nothing.

My prospect received an email with the call report.

That is the content of @notify_team_by_email, meant for your team only: this action refuses the contact's address. To write to them, use @send_email.

@send_email was refused.

The contact has no known email address and gave none during the call, or the requested address is not theirs. Add the email to the campaign data, or have the assistant ask for it during the call and collect it (Collect information tab).

My Slack or Teams action does not show up when I type @.

It is offered once the workspace is connected and at least one channel chosen, in the assistant's Team alerts tab.

Do the instructions run for voicemails?

Yes, but only on the last planned attempt: a voicemail on the first attempt triggers nothing, the call is tried again. If an instruction is only about conversations, say so in the condition: "if someone picked up…".