Skip to main content

Support

Troubleshooting

The failures that actually happen, and their causes.

Test calls

The test call never arrives

CauseCheck
Number missing country codeUse E.164. US: +15551234567. India: +919876543210. The dashboard shows a "tested format" preview before it triggers the call.
Trial credit exhaustedSettings → Plan and billing → Usage. New accounts get $25 credit.
Number on the suppression listSettings → Compliance → Do Not Call List. Search for the number.
Carrier rejected the callSome carriers block calls from new numbers, more often internationally. Try a different from-number.
Handset spam filterApple and Google filters drop some calls silently with no failure signal on your side. Try a different from-number.

The call connects but the Finn says nothing

The most common cause is a welcome message set to defined with an empty field. Either put text in the field or switch to dynamic. See agents-prompting.

Other causes: the voice is not assigned, or is assigned to a voice that requires a region match (agents-voice). If you changed AI service configuration, reset it to default.

Building a Finn

Save is grayed out

A required field is empty (usually Name, Voice, or Language) or a workflow node has a validation error, shown as a red border. If neither applies, your browser session expired. Refresh and log back in. Drafts autosave, so edits survive.

The Finn ignores your script

CauseFix
Identity field too vague"You are Maya, a friendly receptionist" beats "You are an AI assistant."
Guardrails buried among other instructionsMove them into Style and Guardrails on their own.
Fields contradict each otherAudit Identity, Style, Welcome, and Workflow together.
Knowledge base holds contradictory source materialOpen the Knowledge tab in the Data Extractor Sidebar for a sample call, see what was retrieved, delete the contradiction.
You need a fixed order and are using free-form promptingUse a workflow instead.

The Finn asks the same question twice

The caller's answer did not match what the Ask node expected, so it asked again. Soften the parser or rephrase the question. If the answer was captured but not remembered, add to Response Guidelines: "Once a question is answered, do not ask it again." Also check the workflow for an edge looping back into the Ask node.

Pacing is wrong

SymptomFix
Too talkyResponse Guidelines: "Keep responses under two sentences unless asked for more detail."
Too briefResponse Guidelines: "Give complete answers with examples when relevant."
Interrupts the callerLower Settings → Call Settings → Interruption Sensitivity.
Long silencesLower max_idle_duration.
Calls cut off around five minuteslimit_call_duration defaults to 5 minutes. Raising it also raises the per-call cost ceiling.

Knowledge base

SymptomCause
Answers missing from indexed documentsIndexing not finished. Knowledge Base tab should read Indexed. Click Re-index Now if it is stuck.
Scanned PDF returns nothingImage-only PDFs are not OCR'd. Convert to text before uploading.
Retrieval misses a section of a long documentSplit into smaller focused files.
Upload rejectedFiles over 25 MB are refused. Compress or split.
Answers are outdatedDelete the old file and upload the new one. For website imports, use Re-crawl Now.

When retrieval looks wrong, do not guess. Open the Knowledge tab in the Data Extractor Sidebar for a real call and read the snippets that came back. See agents-knowledge-base.

Deployments

Launched, but no calls go out

CauseCheck
Outside the calling-hour windowDeployments dial only inside the configured window, in the caller's local time. Launching at 8 AM PT with 9-6 hours means nothing dials until 9.
Audience is emptyOpen the audience and confirm rows exist.
Every number suppressedSettings → Compliance → Do Not Call List.
From-number not registeredUS numbers without caller ID registration fail silently. Settings → Phone numbers → Registration Status.
Spend cap reachedThe health panel on Live Deployments shows "Cap reached."
Deployment pausedLive Deployments → open the deployment → status.

See outbound-campaigns and phone-numbers.

Everyone hangs up in the first seconds

Unregistered US caller ID gets labeled SPAM LIKELY by carriers, and answer rates collapse. Register it. Beyond that, generic openings get hung up on. Name yourself, name a company the caller recognizes, and state a reason they would expect. If the list is cold, low pickup-and-stay is a list problem, not a configuration problem.

Pause reasons

Live Deployments → open the deployment → Pause Reason.

ReasonMeaning
Spend cap reachedPer-deployment cap hit. Raise it or close the deployment.
Audience exhaustedAll contacts called, retries included.
Buffer drainedDispatcher ran out of work. Usually self-recovers within a minute.
Manual pauseSomeone hit Pause. Check the audit log.
Carrier outageProvider issue. Resumes when the outage ends.
Plan limit hitMonthly call allotment used.

Pausing does not kill in-flight calls. They finish. Pause stops new dialing only.

Analytics

Zero calls shown when calls happened. Usually the deployment dropdown is on a different deployment. If a "Cached" badge is visible, click Refresh. Calls under 5 seconds are filtered out of some metrics as no-answer, so check the full log in call-logs.

Post-call analysis fields blank. Fields only run against calls that completed after the field was added. Existing calls stay blank permanently. Vague questions also extract poorly: "Did the caller express interest?" works, "Interest level?" does not. See post-call-analysis.

Sentiment looks wrong. Sentiment is an indicator, not a verdict. If it misclassifies consistently, add a custom post-call field with a specific question and use that instead.

Phone numbers

SymptomCause
Bought a number, cannot use itKYC verification pending. India, France, and Germany require it. Settings → Phone numbers → Compliance Status.
Outbound fails with "carrier rejected" (US)Caller ID not registered.
BYO trunk not connectingSIP configuration. See byoc-sip.

Still stuck

Email [email protected] with the deployment ID, the call ID if there is one, and what you expected versus what happened. See support.