* fix: cal ai email * fix: remove * fix: org * replace Cal AI with Cal.ai * fix: use * fix: feedback * fix: types * fix: types * fix: types * fix: tests * Merge branch 'main' into fix/cal-ai-credits * refactor: feedback * refactor: imporvement * fix: type * refactor: feedback * fix: tests * fix: use pbac --------- Co-authored-by: CarinaWolli <wollencarina@gmail.com>
260 lines
8.4 KiB
TypeScript
260 lines
8.4 KiB
TypeScript
import dayjs from "@calcom/dayjs";
|
|
import { getCalEventResponses } from "@calcom/features/bookings/lib/getCalEventResponses";
|
|
import { createDefaultAIPhoneServiceProvider } from "@calcom/features/calAIPhone";
|
|
import { handleInsufficientCredits } from "@calcom/features/ee/billing/helpers/handleInsufficientCredits";
|
|
import { FeaturesRepository } from "@calcom/features/flags/features.repository";
|
|
import { checkRateLimitAndThrowError } from "@calcom/lib/checkRateLimitAndThrowError";
|
|
import logger from "@calcom/lib/logger";
|
|
import prisma from "@calcom/prisma";
|
|
import { CreditUsageType } from "@calcom/prisma/enums";
|
|
|
|
interface ExecuteAIPhoneCallPayload {
|
|
workflowReminderId: number;
|
|
agentId: string;
|
|
fromNumber: string;
|
|
toNumber: string;
|
|
bookingUid: string;
|
|
userId: number | null;
|
|
teamId: number | null;
|
|
providerAgentId: string;
|
|
}
|
|
const log = logger.getSubLogger({ prefix: [`[[executeAIPhoneCall] `] });
|
|
|
|
export async function executeAIPhoneCall(payload: string) {
|
|
let data: ExecuteAIPhoneCallPayload;
|
|
try {
|
|
data = JSON.parse(payload);
|
|
} catch (error) {
|
|
log.error("Failed to parse AI phone call payload", { error, payload });
|
|
throw new Error("Invalid JSON payload");
|
|
}
|
|
|
|
const featuresRepository = new FeaturesRepository(prisma);
|
|
const calAIVoiceAgents = await featuresRepository.checkIfFeatureIsEnabledGlobally("cal-ai-voice-agents");
|
|
|
|
if (!calAIVoiceAgents) {
|
|
log.warn("Cal.ai voice agents are disabled - skipping AI phone call");
|
|
return;
|
|
}
|
|
|
|
log.info(`Executing AI phone call for workflow reminder ${data.workflowReminderId}`, data);
|
|
|
|
try {
|
|
// Check if the workflow reminder still exists and is scheduled
|
|
const workflowReminder = await prisma.workflowReminder.findUnique({
|
|
where: { id: data.workflowReminderId },
|
|
select: {
|
|
id: true,
|
|
scheduled: true,
|
|
referenceId: true,
|
|
workflowStep: {
|
|
select: {
|
|
agent: {
|
|
select: {
|
|
outboundPhoneNumbers: {
|
|
select: {
|
|
phoneNumber: true,
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
booking: {
|
|
select: {
|
|
uid: true,
|
|
startTime: true,
|
|
endTime: true,
|
|
eventTypeId: true,
|
|
responses: true,
|
|
location: true,
|
|
description: true,
|
|
attendees: {
|
|
select: {
|
|
name: true,
|
|
email: true,
|
|
phoneNumber: true,
|
|
timeZone: true,
|
|
},
|
|
},
|
|
eventType: {
|
|
select: {
|
|
title: true,
|
|
bookingFields: true,
|
|
},
|
|
},
|
|
user: {
|
|
select: {
|
|
name: true,
|
|
timeZone: true,
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
});
|
|
|
|
if (!workflowReminder || !workflowReminder.scheduled) {
|
|
log.warn(`Workflow reminder ${data.workflowReminderId} not found or not scheduled`);
|
|
throw new Error("Reminder not found or not scheduled");
|
|
}
|
|
|
|
if (data.userId || data.teamId) {
|
|
const { CreditService } = await import("@calcom/features/ee/billing/credit-service");
|
|
const creditService = new CreditService();
|
|
|
|
const hasCredits = await creditService.hasAvailableCredits({
|
|
userId: data.userId || undefined,
|
|
teamId: data.teamId || undefined,
|
|
});
|
|
|
|
if (!hasCredits) {
|
|
log.warn(`Insufficient credits for AI phone call for workflow reminder ${data.workflowReminderId}`, {
|
|
userId: data.userId,
|
|
teamId: data.teamId,
|
|
workflowReminderId: data.workflowReminderId,
|
|
bookingUid: workflowReminder.booking?.uid,
|
|
});
|
|
|
|
await handleInsufficientCredits({
|
|
userId: data.userId,
|
|
teamId: data.teamId,
|
|
creditFor: CreditUsageType.CAL_AI_PHONE_CALL,
|
|
prismaClient: prisma,
|
|
context: {
|
|
workflowReminderId: data.workflowReminderId,
|
|
bookingUid: workflowReminder.booking?.uid,
|
|
},
|
|
});
|
|
|
|
return;
|
|
}
|
|
}
|
|
|
|
const rateLimitIdentifier = data.teamId
|
|
? `ai-phone-call:team:${data.teamId}`
|
|
: data.userId
|
|
? `ai-phone-call:user:${data.userId}`
|
|
: null;
|
|
|
|
if (!rateLimitIdentifier) {
|
|
log.warn(`No rate limit identifier found for AI phone call. This should not happen.`, {
|
|
userId: data.userId,
|
|
teamId: data.teamId,
|
|
workflowReminderId: data.workflowReminderId,
|
|
});
|
|
throw new Error("No rate limit identifier found for AI phone call. This should not happen.");
|
|
}
|
|
|
|
// TODO: add better rate limiting for AI phone calls
|
|
if (rateLimitIdentifier) {
|
|
await checkRateLimitAndThrowError({
|
|
rateLimitingType: "core",
|
|
identifier: rateLimitIdentifier,
|
|
});
|
|
}
|
|
|
|
const booking = workflowReminder.booking;
|
|
if (!booking) {
|
|
log.warn(`No booking found for workflow reminder ${data.workflowReminderId}`);
|
|
throw new Error("No booking found");
|
|
}
|
|
|
|
const numberToCall = data.toNumber;
|
|
|
|
if (!numberToCall) {
|
|
log.warn(`No phone number found for attendee in booking ${booking.uid}`);
|
|
throw new Error("No phone number found for attendee");
|
|
}
|
|
|
|
const attendee = booking.attendees[0];
|
|
const timeZone = booking.user?.timeZone || attendee?.timeZone || "UTC";
|
|
|
|
const { responses } = getCalEventResponses({
|
|
bookingFields: booking.eventType?.bookingFields ?? null,
|
|
booking: {
|
|
...booking,
|
|
customInputs: null,
|
|
},
|
|
});
|
|
|
|
const attendeeNameWords = attendee?.name?.trim().split(" ") || [];
|
|
const attendeeNameWordCount = attendeeNameWords.length;
|
|
const attendeeFirstName = attendeeNameWords[0] || "";
|
|
const attendeeLastName = attendeeNameWordCount > 1 ? attendeeNameWords[attendeeNameWordCount - 1] : "";
|
|
|
|
const dynamicVariables = {
|
|
EVENT_NAME: booking.eventType?.title || "",
|
|
EVENT_DATE: dayjs(booking.startTime).tz(timeZone).format("dddd, MMMM D, YYYY"),
|
|
EVENT_TIME: dayjs(booking.startTime).tz(timeZone).format("h:mm A"),
|
|
EVENT_END_TIME: dayjs(booking.endTime).tz(timeZone).format("h:mm A"),
|
|
TIMEZONE: timeZone,
|
|
LOCATION: booking.location || "",
|
|
ORGANIZER_NAME: booking.user?.name || "",
|
|
ATTENDEE_NAME: attendee?.name || "",
|
|
ATTENDEE_FIRST_NAME: attendeeFirstName,
|
|
ATTENDEE_LAST_NAME: attendeeLastName,
|
|
ATTENDEE_EMAIL: attendee?.email || "",
|
|
NUMBER_TO_CALL: numberToCall,
|
|
ATTENDEE_TIMEZONE: attendee?.timeZone || "",
|
|
ADDITIONAL_NOTES: booking.description || "",
|
|
EVENT_START_TIME_IN_ATTENDEE_TIMEZONE: dayjs(booking.startTime)
|
|
.tz(attendee?.timeZone || timeZone)
|
|
.format("h:mm A"),
|
|
EVENT_END_TIME_IN_ATTENDEE_TIMEZONE: dayjs(booking.endTime)
|
|
.tz(attendee?.timeZone || timeZone)
|
|
.format("h:mm A"),
|
|
// DO NOT REMOVE THIS FIELD. It is used for conditional tool routing in prompts
|
|
eventTypeId: booking.eventTypeId?.toString() || "",
|
|
// Include any custom form responses
|
|
...Object.fromEntries(
|
|
Object.entries(responses || {}).map(([key, value]) => [
|
|
key
|
|
.replace(/[^a-zA-Z0-9 ]/g, "")
|
|
.trim()
|
|
.replaceAll(" ", "_")
|
|
.toUpperCase(),
|
|
value.value?.toString() || "",
|
|
])
|
|
),
|
|
};
|
|
|
|
const aiService = createDefaultAIPhoneServiceProvider();
|
|
|
|
await aiService.updateToolsFromAgentId(data.providerAgentId, {
|
|
eventTypeId: booking.eventTypeId,
|
|
timeZone: attendee?.timeZone ?? "Europe/London",
|
|
userId: data.userId,
|
|
teamId: data.teamId,
|
|
});
|
|
|
|
const call = await aiService.createPhoneCall({
|
|
fromNumber: data.fromNumber,
|
|
toNumber: numberToCall,
|
|
dynamicVariables,
|
|
});
|
|
|
|
log.info("AI phone call created successfully:", call);
|
|
|
|
// Update the workflow reminder with the call reference
|
|
await prisma.workflowReminder.update({
|
|
where: { id: data.workflowReminderId },
|
|
data: {
|
|
referenceId: call.call_id,
|
|
scheduled: true,
|
|
},
|
|
});
|
|
|
|
log.info(`AI phone call executed successfully for workflow reminder ${data.workflowReminderId}`, {
|
|
callId: call.call_id,
|
|
agentId: data.agentId,
|
|
fromNumber: data.fromNumber,
|
|
toNumber: numberToCall,
|
|
bookingUid: data.bookingUid,
|
|
});
|
|
} catch (error) {
|
|
log.error("Error executing AI phone call:", error);
|
|
throw error;
|
|
}
|
|
}
|