Call screening API: know the caller before you answer
A call screening API that tells your stack who is calling while the phone rings. Voice AI agents stop paying tokens and minutes for scam callers. UCaaS platforms sell the screening as a feature. Support teams only pick up real calls.
Great Lakes Freight Co.
VERIFIED · Freight · Chicago, US
+1 888 221 1161
SPAM · score 100 · 43 complaints
+44 7476 937994
UNKNOWN · no complaint history
One GET per call. Your rules decide the action.
1
GET request while the phone rings. The caller never notices the check.
195
countries of business identity, so verified callers get greeted by name.
60s
between spam data updates. A robocall wave that started at 9:00 is dropped at 9:01.
Your AI answers every call. That is the problem.
A spam call to a voice agent is not free. It burns speech-to-text, LLM tokens, telephony minutes and a concurrency slot on a caller who was never a customer. A robocall campaign multiplies that by thousands in an afternoon.
Screening happens before the pickup, so the meter never starts. For repeat offenders, pair it with the spam call blocking API and drop them outright.
| One spam call | Unscreened | Screened |
|---|---|---|
| Speech-to-text and LLM tokens | wasted | never started |
| Telephony minutes | wasted | zero |
| Concurrency slot | blocked | free for a customer |
| Human agent time | interrupted | untouched |
| Cost of the check | none | one GET |
Screening rules you write yourself
The API returns facts, not opinions. Your routing logic stays yours, in your stack.
{
"entity_type": "UNKNOWN",
"reputation": "SPAM",
"is_spam": true,
"spam_score": 100,
"total_complaints": 43,
"complaints": [{
"Subject": "Calls pretending to be government",
"RecordedMessageOrRobocall": "Y"
}]
}if spam_score > 80 → drop
Known spammers and robocallers never reach an agent, human or AI. The complaint subjects tell you what the campaign was.
if reputation == VERIFIED → greet by name
Business identity from 195 countries lets the agent open with context instead of an interrogation.
if entity == UNKNOWN → voicemail
Unknowns get a softer path: voicemail, a callback flow, or extra verification before a human's time is spent.
Who screens calls, and what they get out of it
Three kinds of teams, three different wins from the same request.
Voice AI platforms
Every answered spam call is a token bill, a minute bill and a slot your real customer could have had. Screen before the pickup and the model only ever talks to people who might buy. A fraud campaign against one Voice AI CPaaS customer was dropped before it reached their agents or burned their minutes.
Works from any bot framework through plain REST.
UCaaS and CPaaS
Screening is a feature your customers pay for. Named callers and spam labels on every seat, sold as a premium tier under your brand.
Support and sales teams
Agents see the company before they say hello, and robocalls never ring the queue. Handle time drops, morale does not.
Already on Asterisk, Aircall, Dialpad or GoHighLevel?
The native Asterisk module and the turnkey Aircall, Dialpad and GoHighLevel integrations run the same screening without a line of code. Everything else calls the REST API from the dialplan or middleware.
Screening your own numbers?
Check one number with the free caller ID test, or monitor the whole outbound inventory with the DID reputation monitoring API, which tells you exactly where a number is flagged.
Call screening API questions
Screen your next call
Plans from $25 a month. Voice AI platform with real volume? Bring the numbers and we will quote the tier.