Currently building a demo AI phone system, an immigration law firm with offices in Orlando, FL and Midvale, UT. The problem we're solving: They get calls in three languages — English, Spanish, and Haitian Creole. Callers are dealing with sensitive immigration situations. Missed calls mean missed clients. Phone intake is slow and stressful for everyone. What we're building: A single AI receptionist named Angel that handles every call, in all three languages. How the new caller flow works: Caller says they need help → Angel explains the intake form process → collects their phone number → saves their record → texts them the form link → tells them to call back to book once it's filled out. No long phone intakes. No repeating sensitive details to a receptionist. They fill it out on their own time. Returning caller flow: Caller says they've done the form → Angel looks them up by phone → greets them by name → checks live calendar availability → books the consultation on the call. Also handles: - Rescheduling and canceling appointments - Taking messages after hours - Logging questions for the team to follow up on - Automatic end-of-call reports Tech stack: - VAPI with GPT-4.1 Mini - 11 n8n workflows connected through MCP - Google Sheets CRM - Google Calendar for availability - Twilio SMS for the intake form - Trilingual language detection Still in demo — testing and refining before we hand it over. But the core flow is working. This is the kind of system we build at DeskWolf for service businesses that are missing calls or spending too long on intake. Questions? Drop them below.