Compliance

Designed for real phone rules—not workarounds.

When someone hears an AI voice on a phone call, U.S. rules generally treat that like other automated or recorded voices. A natural-sounding conversation does not remove those rules. We build the product around that reality.

Not legal advice. This page describes product design and responsibility split. It is not a legal opinion. Live calling programs need qualified counsel. We do not certify that your consent language or exemption is legally sufficient—we enforce the policy you tag on campaigns.

One plain product rule

If the callee hears synthetic / AI speech, treat the program like other artificial or prerecorded voice traffic for planning purposes: purpose, consent or valid basis, disclosures, opt-out, and records matter. “It was an interactive LLM” is not a product posture we sell as a free pass.

Who owns what

Aligned with our Terms: you are the calling entity; we are the control plane.

Outgoing Voice

What the software enforces

→ Campaign purpose and destination-class allowlists
→ Lawful-basis tags and “consent artifact required” when you enable them
→ Pre-dial gate: suppressions, frequency caps, campaign on/off
→ Audit snapshot of the rules used on each attempt (pass or block)
→ Forced open flags: identity + AI disclosure (+ optional paid-for-by style language)
→ Opt-out tools → suppression so redials don’t pass the gate
→ No voice clones of real people (brand AI voices only)
→ Dry-run testing before you spend on live placement

You (calling entity)

What you still own

→ Being the organization that places the call (“maker of the call”)
→ Where numbers come from and why you’re allowed to call them
→ Consent wording and artifacts that cover AI / artificial voice where required
→ Exemptions you claim (we store tags; we don’t litigate them for you)
→ State rules (hours, recording consent, local requirements)
→ Election, charity, healthcare, or industry rules when they apply
→ Staffing human callbacks when the agent escalates
→ Counsel review before large live programs

What the pre-dial gate actually does

Concrete stops—not a vague “AI compliance” badge. Blocked attempts are stored with a reason.

If this is true…	What happens
Campaign is disabled	No dial. Logged as campaign disabled.
Number is on your suppression list	No dial. Opt-outs and manual blocks stay blocked.
Destination class not allowed (e.g. wireless on a landline-only campaign)	No dial. Mismatch recorded.
Basis is consent-type and no consent artifact is asserted	No dial. Consent required for that tag set.
Frequency cap exceeded for that number	No dial. Cap from campaign or basis conditions.
Lawful basis is explicitly BLOCK	No dial. Policy says do not call.
All tagged checks pass	Allow. Snapshot of rules kept for audit / dry-run / live attempt.

Try it without a live call: Test a number in the app.

Myths we won’t sell

“AI chat means TCPA doesn’t apply.”

If they hear AI voice, plan for artificial/prerecorded-style rules—not a free pass.

“It’s political, so anything goes.”

Politics is not a product exemption we market. Consent and purpose still matter.

“Outgoing Voice gets opt-ins for us.”

You bring the relationship and lawful basis. We enforce tags and suppressions.

“Clone the doctor / candidate for trust.”

Hard no. Brand synthetic voices only—no real-person clones.

Safer defaults for pilots

Start simple

Appointment confirms first: clear purpose, people you already know, easy to measure containment and no-shows.

No voice clones

Only brand AI voices you choose. Pretending to be someone real is banned in the product.

Stop when unsure

Missing permission tags or a blocked number means no dial. Dry-run before live volume.

Also see Terms of Service (customer indemnity for calling programs), Privacy Policy, and Product modules.

Next steps

See the modules, walk the journey, or talk through an appointment-confirm pilot with counsel in the loop.

Product How it works Test a number Talk about a pilot