Question 1

What is Voxguard?

Accepted Answer

Voxguard is the QA and audit layer for customer-facing AI agents. It observes prompt-and-response pairs, scores them against your rulebook, and records findings before an off-brand, unsupported, or out-of-policy answer becomes a customer or regulatory problem.

Question 2

Who is Voxguard for?

Accepted Answer

Voxguard is built for operations, legal, CX, and product leaders shipping AI into chat, voice, or sales channels. It is especially useful for mid-market and enterprise teams that need an auditable control layer around a live agent without replacing their model or inference stack.

Question 3

What does Voxguard audit?

Accepted Answer

Voxguard audits hallucination and unsupported claims, brand-voice drift, disallowed pricing or medical claims, PII spill, and regulatory gates. Your team can add custom rules so the audit reflects your product, policies, evidence sources, and risk appetite.

Question 4

How do scoring and severity work?

Accepted Answer

Each turn is scored against the configured rulebook and produces a finding when it breaks a rule. Severity reflects the likely customer, operational, or regulatory impact and sets the response urgency: critical findings can page the team immediately, while the weekly report groups lower-severity patterns for review and remediation.

Question 5

Which regulations does Voxguard map to?

Accepted Answer

Voxguard provides rulebook gate sets for the EU AI Act, including Article 50 transparency requirements, Colorado SB 24-205 (the Colorado AI Act), and New York City Local Law 144. The rulebook is designed to support operational readiness and evidence collection; it does not replace advice from your legal counsel.

Question 6

How does onboarding work?

Accepted Answer

We begin with your agent surface, risk priorities, and existing policy or evidence sources. Voxguard then receives the prompt-and-response tuple through an SDK-injected forwarder, SIEM hop, or webhook, runs the agreed rulebook in your region, and returns findings to the control plane. Teams can start with a focused audit of a recent week of conversations before expanding coverage.

Question 7

What pricing tiers are available?

Accepted Answer

Starter is $1,000 per month for one workflow and up to 25,000 scored turns. Growth is $2,500 per month for up to 100,000 scored turns, custom rules, real-time escalation, and the regulation gate sets. Enterprise is custom-scoped for multiple agents, regions, and rulebooks, starting around 250,000 scored turns per month. There is no per-call tax.

Question 8

How can I contact the Voxguard team?

Accepted Answer

Email voxguard@polsia.app to request an audit slot, ask a product question, or discuss the right tier for your AI surface. The team can also start with a 30-minute anonymised audit of a recent week of conversations.