Frozen, Not Silent
Explain frozen accounts without tipping off: safe words for every agent and bot.
The job
Stops agents and bots from tipping off suspects, while also stopping genuine customers with frozen accounts from being stonewalled into repeat calls and complaints to the regulator.
The moment
In a review, the money laundering reporting officer replays last week's bot conversations about frozen accounts, marking risky wording in red.
23 conversations where the bot told customers their account was "flagged for money laundering" are struck through live. The officer approves a safe script before the meeting ends.
The same red pen works on agents' drafts. A care agent's draft saying the account was under money laundering review was flagged and rewritten to "your account needs a routine check; we'll update you by Tuesday".
The next week no risky phrases appear, and repeat calls about frozen accounts are down by a third.
What it does
- Approve a safe script
- Publish to agent macros
- Publish to the AI agent's knowledge / switch on its guardrail
- Send a coaching list to QA
- Send an approved customer update
- Request an expedited review / refer the red lane
What you see
Three-lane review (red, grey, green) with a live red-pen strike-through on drafts and bot replays; restricted screen, never on a wall
What it moves
Tipping-off phrases in agent and bot replies each week, with a target of zero, repeat contacts per customer with a frozen account, and mentions of complaints to the regulator.
Built for
- Operations
- CX leaders & VoC
- Contact-centre operations
- Frontline staff
