In-call education
In-call education is an optional, per-campaign feature that lets the AI voice break character during the call and deliver a short security-awareness debrief the moment a target either falls for the pretext or clearly detects and refuses it. It turns the simulation into a teachable moment at the point of maximum salience, rather than waiting for a follow-up email after the call ends. It is off by default, and the configured post-call follow-up (see Follow-up) and recorded outcome still happen exactly as before, regardless of whether a debrief was delivered.
Every debrief, in every mode, always ends with a non-removable compliance disclosure: it discloses that this was an authorized test run by the target's own organization, states that nothing real happened and no data was collected, stays brief (under about 90 seconds), and ends politely. This floor supports EU AI Act Article 50 disclosure obligations and cannot be turned off or edited by an operator, in any mode.
Prerequisites
- A campaign in the builder, at the Post-simulation training step (the "On the call" card lives inside the outcome drawer's Answered section, it only applies to calls the target actually answers).
- To customize the org-wide educator identity used by the Callstrike Educator handoff, admin access to Settings → Compliance → In-call Education.
How to use it
- In the campaign builder's Post-simulation training step, open the Answered outcome drawer and find the Educate on the call card.
- Pick one of four modes (see the table below). The default is Off.
- Depending on the mode, fill in the persona fields, a custom educator prompt, or the two debrief scripts, or leave them blank to use the shipped/org defaults.
- Save the campaign. The compliance floor and a live preview of the handoff are always shown beneath the mode-specific fields.
- Optionally, set org-wide defaults once under Settings → Compliance → In-call Education (educator name, security team name, CISO name, company name, tone, voice), any campaign field left blank inherits these instead of the shipped defaults.
Configuration & options
| Mode | What happens on the call | Configurable |
|---|---|---|
| Off (default) | No pivot; the call behaves exactly as it does today. | None |
| Callstrike Educator | The attacker assistant hands the call off to a dedicated, transient educator assistant, contextual, bounded to about 75 seconds, the instant the target complies or clearly detects/refuses. | Persona fields: company name, security team name, CISO name, educator name, tone. Blank fields inherit the workspace's org-wide defaults. |
| Custom Educator | Same handoff mechanism as Callstrike Educator, but the educator's prompt body is your own free text instead of the persona-assembled brief. | A free-text educator prompt (the compliance floor is still appended automatically, last, and verbatim). |
| Static Script | The attacker assistant itself reads a fixed, pre-reviewed script; no handoff to a second assistant. | Two editable scripts: one for "target complies," one for "target detects it." Pre-filled with sensible defaults you can edit or leave as-is. |
Switching modes clears fields that don't apply to the new mode (for example, leaving Custom Educator drops the custom prompt so it can't linger on a Static Script or Callstrike Educator config).
Gotchas & limitations
- The compliance floor is never editable. In every mode, the disclosure, no-data-collection statement, brevity cap, and polite close are appended automatically, last, regardless of what an operator writes in a persona field or custom prompt.
- The recorded outcome doesn't change. Whether a debrief was spoken or not, the call's outcome (phished / detected) is still resolved from the same post-call success evaluation used elsewhere, the debrief's reason is recorded as context alongside it, not as the source of truth.
- Only applies to answered calls. The card lives in the outcome drawer's Answered section; there is no in-call debrief path for calls that aren't answered.
- The debrief fires at most once per call. A malformed or duplicate tool invocation is handled safely and does not produce a second timeline entry or break the call.
- Static Script vs. the handoff modes use different delivery mechanisms. Static Script has the attacker assistant itself read the script; Callstrike Educator and Custom Educator hand the call off to a separate transient assistant. If a debrief isn't firing as expected, check which mode the campaign is actually set to.
Best practices
- Turn it on for first-time or high-risk targets, correcting the moment it happens, while the call is still fresh, tends to land harder than an email alone.
- Leave it off for covert or repeat campaigns where tipping the target off mid-call would skew results, or when you'd rather debrief everyone quietly by email after the fact.
- Set the org-wide educator defaults once in Settings so individual campaigns can leave persona fields blank and still sound consistent.
Troubleshooting / FAQ
Does the in-call debrief replace the follow-up email? No. The configured post-call follow-up (see Follow-up) still fires after the call regardless of whether a debrief was delivered.
Can I remove the "this was a test" disclosure to make the simulation feel more realistic? No, the compliance floor is appended automatically in every mode and cannot be edited or turned off.
Does enabling this change how a target's compliance is scored? No. The outcome is still resolved from the same success-evaluation logic used everywhere else; the in-call debrief's reason is recorded as additional context, not as the scoring signal.