Document ID: LCR-POL-AI-v1.1 - Effective Date: 19 February 2026
Havio uses AI to power business phone workflows such as answering calls, qualifying callers, booking appointments, creating summaries, and routing calls to humans. This policy explains how AI is used and what customers should control before launch.
1. How Havio Uses AI
| Context | Description |
|---|---|
| AI receptionist workflows | Speech recognition, language understanding, response generation, summarization, tagging, routing, and workflow actions |
| Internal product operations | Drafting, testing, evaluation, support triage, documentation, and engineering assistance |
| Customer configuration | Approved knowledge, scripts, policies, routing rules, fallback contacts, and integrations configured for each customer |
2. Customer Control
Customers should approve the knowledge, scripts, escalation rules, recording/disclosure language, connected tools, and fallback behavior before production use. Havio is designed to use configured business knowledge and workflow rules rather than unrestricted open-ended answers.
3. Voice, Recording, and Disclosure
Customers are responsible for lawful call recording, AI disclosure, and caller notices in the regions where they operate. Havio supports disclosure and consent language in call flows, but the customer controls the business context and must confirm what is required for its callers.
Havio is not intended to identify a person uniquely from their voice unless that feature is explicitly enabled under a written agreement and an appropriate lawful basis has been established.
4. Providers and Subprocessors
Havio may use third-party AI model, telephony, hosting, analytics, CRM, calendar, and support providers to deliver the service. Specific providers may vary by customer configuration, region, integration, and plan. Customer-specific data processing terms or subprocessor lists are available where applicable.
5. Output Accuracy
AI systems can mishear, mistranscribe, summarize incorrectly, or provide an answer outside the customer's expectations. Customers should configure fallback rules, review transcripts, and keep sensitive or regulated decisions with humans unless a reviewed workflow exists.
Havio does not guarantee error-free AI output. We design workflows to reduce risk through approved knowledge, human handoff, review loops, and monitoring.
6. Prohibited Use
Customers may not use Havio for unlawful deception, impersonation, harassment, fraud, spam, unauthorized recording, illegal outreach, or any use prohibited by our Acceptable Use Policy.
7. Changes
AI providers, models, and product capabilities evolve quickly. Havio may update this policy as its architecture or provider stack changes.
Policy Changelog
| Version | Date | Summary |
|---|---|---|
| v1.1 | 2026-06-29 | Updated AI policy for Havio SaaS workflows |
| v1.0 | 2026-02-19 | Initial publication |