AI Features & How We Use AI

AI is used in three main places, all of them user-initiated and transparent:

• Drafting messages and templates — when you describe a message in plain English, the AI drafts SMS, email, and voice variants tuned to each channel.
• Compliance analysis — the AI classifies messages as informational or marketing, flags TCPA risk indicators, and suggests remediation.
• Voice consent calls — the voice AI conducts conversational phone calls to read disclosures and capture verbal consent.

You can also expose Respectly's tools to your own AI agents (ChatGPT, GitHub Copilot, Google Gemini) via our MCP integration on Pro and Enterprise.

Message drafting and compliance analysis use Azure OpenAI GPT-class models in our private Azure environment. Voice calls use UltraVox for real-time conversational speech. We do not train our models on your data and your prompts and content are not shared between tenants.

Before any send, the AI scores the message for TCPA, CAN-SPAM, and CASL risk indicators and surfaces concerns in the UI as risk badges. If a problem is detected (e.g., a marketing message going to a contact without a written opt-in, or a prerecorded-voice message to a wireless number without consent), the engine blocks the send and offers a one-click "fix" that rewrites the content or suggests the consent action you need.

Free — 3/day. Standard — 50/day. Pro and Enterprise — unlimited. The check counter resets daily at midnight UTC.

Pro includes 100 AI voice minutes per month. Enterprise is custom (unlimited available). Free and Standard do not include AI voice minutes; you can upgrade or buy a +100 minutes add-on at $29.

Yes, on Pro and Enterprise only. We expose 40+ Respectly tools via MCP (Model Context Protocol), the open standard that ChatGPT, Claude, GitHub Copilot, Google Gemini, and other agents use to call external services. Pro includes 3 connected agents with 100 messages/day each; Enterprise is unlimited.

The AI is trained on the compliance rules and will tend to produce messages that include sender ID, opt-out instructions, and appropriate language for the channel. However, the final compliance check is independent of the drafting model — every message, AI-generated or hand-written, passes through the same deterministic Policy Routing Engine before sending. Don't rely on the drafting model alone; rely on the routing engine.

No. AI drafts always require your review and explicit "Send" or "Schedule" click. The AI never sends autonomously. Workflows can be configured to send messages without manual review, but the message content in that case is one you authored or approved as a template.