One contract for every contact, from signal to written outcome

The short version

A signal arrives from your system. emable Voice checks whether a contact is allowed, a person or a rule that person set approves it, and just before the contact the case is checked again. The AI says it is an AI, stays within the approved script and hands the case to a person when unsure. Only the operations you allow are written back to your CRM.

1. A signal from your system

A callback request, an approved next best action, a service event or a fault report. The signal carries a reference to the case and only the fields the script needs. When the same signal arrives twice, it becomes one case, not two contacts.

2. Screening before anyone is asked to approve

emable Voice checks the do-not-contact list and opt-outs, consent or another valid reason for this purpose, the contact rules for the channel and the agreed limits. If a check fails, there is no contact.

3. A person, or a rule that person set, approves exactly what will happen

An advisor or manager approves the purpose, the script and the content of the contact. If the content changes after approval, the approval no longer applies and has to be given again. An approval can also expire.

A person can also set rules for an agent: which purposes, hours and daily limits it may use without asking. A contact that fits the rules is approved by the rule; anything else waits for a person. Only a person can raise an agent to that level, never the agent itself.

4. Scheduled within allowed hours

The approved contact waits in a queue for its time window. The number of attempts is limited.

5. Checked again just before the contact

Just before the call or SMS, the case is checked again: is it still open, is the price still valid, is consent still valid, may the number still be contacted, is it still within the allowed hours and limits. If something has changed, the contact is cancelled or moved.

6. The contact itself

• AI disclosure: every call opens with the approved AI disclosure, so the person knows from the first sentence that an AI is speaking.
• Within the script: the AI stays within the approved purpose and script.
• Recipient verification: caller ID is not proof of identity, so the person is verified before any sensitive detail.
• Opt-outs: when someone opts out, that opt-out is checked before every later attempt.

7. Handover or a recorded outcome

When the AI is unsure or a question is out of scope, it ends the call politely and hands the case to the advisor or manager with a summary of the conversation. Otherwise the contact ends with an outcome in the form the script defines, for example a booked meeting or a refusal with its reason.

8. Only allowed operations are written back

The outcome is written to your CRM or core system through an adapter, and only with the operations you allow for that workflow. Your CRM stays the source of truth.

9. Audit and metering at every step

Each step leaves an audit record: who approved what, which checks ran and with what result, what the AI said and which tools it called, what was written and what was handed over. Minutes and SMS are counted per contact and shown on a transparent bill.

What it will not do

• Sign insurance or investment contracts on its own
• Run mass calling or marketing campaigns
• Call a number just because it is public
• Treat caller ID as proof of identity
• Treat a delivery receipt as proof that an SMS was read