Dashboard · AI monitoring · 2026
Sentiox
Independent monitoring for AI phone agents: scheduled test calls, transcript grading and alerts.
The problem
Pharmacies and clinics are putting AI agents on their phone lines, and the only people finding out when those agents hallucinate hours or lose a refill request are the patients.
What we built
We designed the product and built the monitoring dashboard: scenario-based test calls on a schedule, transcript evaluation against the business’s real facts, callback verification that checks whether the promised action actually happened, and an alerting layer with severity, history and rerun.
Highlights
- 01
Overview with uptime, accuracy, response time and an accuracy trend across scenarios.
- 02
Test run detail with the graded transcript and the exact lines that failed highlighted.
- 03
Callback verification: call back later and confirm the refill or booking really landed.
- 04
Alerts with severity, consecutive-failure tracking and one-click rerun.
Inside the product
In one line
Sentiox places automated test calls against a business’s AI phone agent, grades every transcript with an LLM, and alerts the owner the moment the agent gives wrong information or drops a request.