From customer history
to accountable action.

Basis: anonymized internal commercial intelligence project records reviewed for this showcase. This is a TAG AI Solutions use case, prepared September 2026. The 22.35% baseline and ≥30% target are project-record figures; the target is not an achieved result.

1. docs/GOAL.md (August 21, 2026): 78 of 349 qualifying new customers made a second purchase within 60 days; target ≥30% within two quarters of the pilot. The plan calls for a four-week pilot with a comparison group.
2. README.md (status September 9, 2026): implemented commercial capabilities, intentional trial shutdown, and July 31, 2026 data cutoff.
3. asistente/docs/CAPABILITY-VERIFICATION.md (August 27, 2026): action ownership, customer context, outcome verification, coaching, exports, and external data boundaries.
4. Historical conversation evaluation (September 9, 2026): historical review of 329 human turns; 251 clean passes, 65 review cases, 13 failures. Voice quality was a historical weakness. The evaluation is not a test of the September 9 release, and its independent judge could not recompute every business fact.

Measurement limits: no causal revenue lift, retention improvement, productivity gain, or realized return on investment is claimed. Recommendations indicate potential opportunity; verified outcome records do not by themselves establish causation. The comparison-group design above is the proposed route to stronger evidence.

Scope: product and operating-model showcase. No private customer transaction records, customer identities, credentials, or client endorsements are embedded. Diagram labels describe implemented workflows or explicitly proposed measurement steps.