Top 75%
in this catalogue (796 scored practices)
Scores cluster high, so position within the catalogue is often more telling than the number alone.
Evidence of impact / measured public value2/3
Consistent pre/post figures (60% resolution, 43,000+ served, 99% adherence vs. a stated ~50% baseline) are reported by TGR and its vendor across two independent Chilean outlets, but there is no independently published evaluation and only self-reported operational metrics from a single agency.
Transparency, fairness & accountability1/3
TGR discloses how the agent escalates complex or sensitive cases to a human with full context, but no public evaluation report, error-rate data, or algorithmic transparency register for the system has been published.
Transferability / demonstrated replication1/3
The same platform, built with vendor CGS Nexus, continued into a second tax season (Operación Renta 2026), but no other Chilean agency or country has yet been reported adopting it.
Scalability beyond pilot2/3
The agent moved from a second-half-2024 build directly into full operational use across an entire national tax season, serving over 43,000 callers, and continued into a second season in 2026.
Governance, capability & sustainability1/3
Design includes automatic escalation to human agents with preserved context for sensitive or complex cases, but no independent oversight body, audit, or published risk assessment is described in available sources.
Evidoria. TGR's AI Voice Agent — Chile's Treasury Resolves 60% of Tax-Season Calls Without a Human. Persistent ID: 3ee87dc1-083c-446d-9d83-af6ca05094d4.