What do the analysis run states mean?
Callmine runs move through queued, running, completed, failed, cancelled, and no_match. Heartbeats keep the live view connected across proxies.
TL;DR
Runs move through six states: queued, running, completed,
failed, cancelled, and no_match. The live view at
/runs/<jobId> streams progress and auto-reconnects if the
connection drops. State transitions are server-authoritative —
closing your browser doesn't pause anything.
The states
- ·
queued— Callmine has accepted the request and is waiting for a worker slot. Concurrent-job limits are per-plan; queued runs pick up as earlier runs complete. - ·
running— Active. The live view streams stage-by-stage progress: Gong fetch, transcript pull, per-call LLM, executive summary, completion. - ·
completed— Finished successfully. The report is available at/reports/<reportId>. The live view auto-navigates there. - ·
failed— Unrecoverable error during execution. Common causes: Gong credential rejected mid-run, LLM provider returned a permanent error, an internal bug. Emailhello@callmine.aiwith the/runs/<jobId>URL. - ·
cancelled— You or another admin cancelled the run from/runs/<jobId>. Partial state is preserved but no report is generated. - ·
no_match— Filters returned zero matching calls before any LLM work. No quota consumed.
Live progress
The live view streams progress over a long-lived connection that auto-reconnects if your network drops or a proxy times it out. The run keeps progressing server-side regardless of whether you are watching — closing the tab does not pause anything.
Reopen /runs/<jobId> to resume watching, or check /dashboard
for the final report once it's posted.
When a run consumes quota
Quota is consumed atomically per analyzed call at completion of the
LLM stage. A run that fails mid-batch consumes quota for the calls
that already finished — not for calls that never started. A
no_match or pre-LLM cancel consumes zero.