AI Transparency Notice

This website

The public Haultro marketing website does not provide an AI chatbot, generate personalized content, score visitors, or make employment, credit, housing, insurance, health, legal, or other consequential decisions. The ROI calculator performs user-controlled arithmetic using disclosed assumptions. Optional analytics measure site use only after consent.

Haultro product features

Haultro deployments may include AI-assisted routing, forecasting, anomaly detection, summaries, classifications, alerts, recommendations, retrieval, voice, image analysis, or workflow support. Exact features, data sources, provider, permitted purposes, retention, and deployment boundary are defined by the customer configuration and written agreement.

Voice calls with the public

When the Haultro voice agent answers a call, it identifies itself as automated at the start of the call. A caller can ask for a person at any time, and each operating customer configures where that handoff lands, such as a dispatcher, a service representative, or a scheduled callback. The agent captures consent where the workflow requires it, and call recording follows the operating customer's notice and consent obligations described in the Acceptable Use Policy. Synthesized voice used on calls is disclosed through the same automated-system identification.

Automated outputs and human review

Some Haultro outputs affect identifiable people or their commercial terms: dynamic pricing quoted on live calls, driver assignment and safety coaching, credit recommendations in SLA workflows, and retention tiering. These outputs run behind published guardrails, including human review before a work order is created, a margin floor on quotes, and logged, overridable actions, and they are maintained against the automated decision-making and consequential-decision definitions in CPPA ADMT regulations (obligations for existing uses begin January 1, 2027) and the Colorado AI Act. A person affected by one of these outputs can request human review through the operating customer that runs the deployment, or by contacting privacy@trunnion.ai; requests are routed to the operator controlling that deployment. Website-form information is never used for consequential automated decisions.

Human review and control

AI output can be incomplete, inaccurate, biased, outdated, or unsuitable. A qualified user must review outputs before action when safety, employment, legal rights, regulatory compliance, financial reporting, environmental reporting, or another material consequence is involved. Deployments should define permissions, approval points, correction paths, logging, monitoring, escalation, and the ability to disable applicable AI features.

Data handling and model training

Customer operational data, call audio, and transcripts are not used to train shared or cross-tenant Haultro or Trunnion models; learning features operate inside the customer's own tenant only, and a customer can further restrict tenant-scoped processing through its written agreement. For third-party providers, whether a model receives data, where processing occurs, whether the provider retains data, and whether provider training is disabled must be confirmed before production use. Do not enter sensitive or confidential information unless the written scope and deployed controls expressly permit it. See the sensitive data section of the Security page for how precise location and voice audio are handled.

Monitoring and reporting

Customers remain responsible for assigning human owners, validating configuration, monitoring outcomes and drift, reviewing logs, handling complaints and corrections, and completing applicable impact assessments. Questions or reports may be sent to contact@haultro.com.