Claim before calling
Reserve the message in a persisted state before the external action begins.
✓ Concurrent workers and restarts see the same delivery authority.DELIVERY RECONCILIATION
Check what happened before sending again.
A provider may accept a request before its response is lost. Nolina holds the request until the outcome can be confirmed.
THE PROBLEM
Automatic retry turns provider ambiguity into duplicate outreach.
Network timeouts, rate limits, and 5xx responses do not prove that no action occurred. Nolina separates proven non-delivery from an unknown outcome.
PRODUCT CONTROLS
Checks before the next action.
Review how each check handles the source, approval, or delivery outcome.
Classify the response
Separate confirmed success, proven pre-action failure, and ambiguous outcome.
✓ Only proven non-delivery can return to an eligible state.Reconcile with the provider
Ask what actually happened instead of inferring from a lost response.
✓ Unknown outcomes remain visible and auditable.Escalate when proof is unavailable
Move irreducible uncertainty to human review.
✓ The system prefers a missed touch to a duplicate touch.Suppressed → stopUnknown send → reconcileAmbiguous reply → review
WHAT TO MEASURE
What to measure in the pilot.
- ✓At-most-once dispatch policy
- ✓Visible uncertain delivery state
- ✓Provider-specific reconciliation
- ✓Human review for irreducible ambiguity
A STRONG FIT WHEN
When this workflow fits your team.
- The sending provider does not guarantee idempotency
- A duplicated invite or email damages trust
- Workers can restart during active delivery
- RevOps needs evidence of what actually happened
RUN A REAL SEGMENT
Test the workflow with your accounts.
Define the target segment, evidence rules, channel, exclusions, and success measure before execution begins.
Request a pilot scope