fix: AI recall-call flow, speech-to-text time parsing, and stale conversation state
- Reset conversation stage to a fresh start on every new voice call (recall and reminder alike), and on every hangup/timeout/goodbye, so an abandoned or completed call never leaves a patient stuck mid-flow for their next call or text. - Add a staleness check to the SMS flow: after a long idle gap, check whether the patient's reply still reads as a continuation before resuming the stored stage, falling back to a friendly "it's been a while" re-greeting when unsure. - Fix parseTime() to handle Twilio's speech-to-text output, which renders "pm"/"am" with periods and often drops the colon in a spoken time (e.g. "100 p.m." -> 1:00 PM) — this was silently failing and causing the AI to loop asking for a time it had already been given. - Extend the voice Gather timeout so the AI doesn't hang up on a patient who just needs a few seconds to respond. - Reword the recall-call opening line and rework the post-"yes" prompt to ask for day and time together. - Fix rclone WebDAV server to serve only the backups folder instead of the entire backend app directory (was exposing .env and source). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
ff
@@ -2,7 +2,7 @@ import twilio from "twilio";
|
||||
import { prisma as db } from "@repo/db/client";
|
||||
import { storage } from "../storage";
|
||||
import { runReminderGraph } from "./reminder-graph";
|
||||
import { runRecallGraph, runRecallBookingStep } from "./recall-graph";
|
||||
import { runRecallGraph, runRecallBookingStep, hasDateOrTimeSignal } from "./recall-graph";
|
||||
import { runNewPatientStep } from "./new-patient-graph";
|
||||
import {
|
||||
runRescheduleStep,
|
||||
@@ -290,11 +290,17 @@ export async function routePatientTurn(params: {
|
||||
if (intent === "yes") {
|
||||
// Patient may have already given a date (or date + time) in this same
|
||||
// reply ("Tuesday 1pm") — try to book right away instead of asking again.
|
||||
// A bare "yes"/"ok" has nothing to parse, so skip straight to asking for
|
||||
// day + time instead of wasting a parse call that used to come back
|
||||
// empty and send a mismatched "Sorry, I didn't catch a day" reply.
|
||||
if (hasDateOrTimeSignal(message)) {
|
||||
const { reply: bookingReply, nextStage: bookingNextStage } = await runRecallBookingStep(
|
||||
message, "recall_asked_datetime", language, patient.id, patient.userId, activeAi.key, activeAi.provider, activeAi.model
|
||||
);
|
||||
return finish(patient.userId, patient.id, [bookingReply], bookingNextStage);
|
||||
}
|
||||
return finish(patient.userId, patient.id, [intentReply], "recall_asked_datetime");
|
||||
}
|
||||
if (intent === "no") {
|
||||
return finish(patient.userId, patient.id, [intentReply], "done");
|
||||
}
|
||||
|
||||
@@ -23,6 +23,19 @@ type GraphStateType = typeof GraphState.State;
|
||||
|
||||
// ── Intent classifier — multilingual yes/no keywords (shared shape with reminder-graph) ──
|
||||
|
||||
/** True if `text` mentions a concrete day/date/time (e.g. "Tuesday", "3/10", "2pm").
|
||||
* Shared by classifyNode (a bare "yes" vs. "yes, Tuesday at 1pm" both count as
|
||||
* intent "yes") and by the caller in patient-conversation-router.ts, which uses
|
||||
* it to decide whether to attempt an immediate booking parse or just ask for a
|
||||
* day/time first. */
|
||||
export function hasDateOrTimeSignal(text: string): boolean {
|
||||
return (
|
||||
/\b(monday|tuesday|wednesday|thursday|friday|saturday|sunday|today|tomorrow|next week)\b/i.test(text) ||
|
||||
/\b\d{1,2}[\/\-]\d{1,2}\b/.test(text) ||
|
||||
/\b\d{1,2}(:\d{2})?\s*(am|pm|a\.m\.|p\.m\.)\b/i.test(text)
|
||||
);
|
||||
}
|
||||
|
||||
function classifyNode(state: GraphStateType) {
|
||||
const text = state.message.toLowerCase().trim();
|
||||
|
||||
@@ -32,10 +45,7 @@ function classifyNode(state: GraphStateType) {
|
||||
// A specific day/date/time, or an appointment-request phrase, is itself a
|
||||
// strong signal the patient wants to book — real replies rarely say a bare
|
||||
// "yes", they jump straight to "Tuesday at 1pm" or "I want an appointment".
|
||||
const mentionsDateOrTime =
|
||||
/\b(monday|tuesday|wednesday|thursday|friday|saturday|sunday|tomorrow|next week)\b/i.test(text) ||
|
||||
/\b\d{1,2}[\/\-]\d{1,2}\b/.test(text) ||
|
||||
/\b\d{1,2}(:\d{2})?\s*(am|pm|a\.m\.|p\.m\.)\b/i.test(text);
|
||||
const mentionsDateOrTime = hasDateOrTimeSignal(text);
|
||||
const mentionsAppointment = /\b(appointment|schedule|book|come in|check.?up|cleaning)\b/i.test(text);
|
||||
|
||||
if (noPatterns.test(text) && !mentionsDateOrTime) return { intent: "no" };
|
||||
@@ -52,13 +62,13 @@ function routeByIntent(state: GraphStateType): string {
|
||||
// ── Fallbacks ──────────────────────────────────────────────────────────────
|
||||
|
||||
const WANTS_BOOKING_FALLBACKS: Record<string, string> = {
|
||||
English: "Great! What day works best for you?",
|
||||
Spanish: "¡Genial! ¿Qué día le viene mejor?",
|
||||
Portuguese: "Ótimo! Qual dia é melhor para você?",
|
||||
Mandarin: "太好了！您哪天方便？",
|
||||
Cantonese: "太好喇！您邊日方便？",
|
||||
Arabic: "رائع! ما هو اليوم الأنسب لك؟",
|
||||
"Haitian Creole": "Bon! Ki jou ki pi bon pou ou?",
|
||||
English: "Great! What day and time work best for you?",
|
||||
Spanish: "¡Genial! ¿Qué día y hora le vienen mejor?",
|
||||
Portuguese: "Ótimo! Qual dia e horário são melhores para você?",
|
||||
Mandarin: "太好了！您哪天、几点方便？",
|
||||
Cantonese: "太好喇！您邊日、幾點方便？",
|
||||
Arabic: "رائع! ما هو اليوم والوقت الأنسب لك؟",
|
||||
"Haitian Creole": "Bon! Ki jou ak ki lè ki pi bon pou ou?",
|
||||
};
|
||||
|
||||
const DECLINES_FALLBACKS: Record<string, string> = {
|
||||
@@ -98,7 +108,7 @@ async function wantsBookingNode(state: GraphStateType, config: any) {
|
||||
{
|
||||
role: "system",
|
||||
content:
|
||||
`You are a friendly dental office assistant. The patient agreed to come in for a routine exam and cleaning. Write a short, warm reply (1-2 sentences max) and ask what day works best for them. Do NOT introduce yourself or say your name — the introduction has already been sent in a separate message. You MUST reply in ${lang}. No formatting, no extra text.`,
|
||||
`You are a friendly dental office assistant. The patient agreed to come in for a routine exam and cleaning. Write a short, warm reply (1-2 sentences max) and ask what day AND time work best for them. Do NOT introduce yourself or say your name — the introduction has already been sent in a separate message. You MUST reply in ${lang}. No formatting, no extra text.`,
|
||||
},
|
||||
{ role: "user", content: `Patient replied: "${state.message}"` },
|
||||
]);
|
||||
|
||||
@@ -392,18 +392,35 @@ async function isSlotAvailable(
|
||||
// ── Time parsing (legacy) ─────────────────────────────────────────────────────
|
||||
|
||||
export async function parseTime(message: string, apiKey: string, provider: AiProvider = "google", model?: string): Promise<string | null> {
|
||||
const t = message.toLowerCase();
|
||||
// Twilio speech-to-text usually renders "pm"/"am" with periods ("p.m."),
|
||||
// and often drops the colon in a spoken time too ("1:00pm" -> "100 pm",
|
||||
// "12:30pm" -> "1230 pm") — normalize both before matching.
|
||||
const t = message.toLowerCase()
|
||||
.replace(/\ba\.m\.?/g, "am")
|
||||
.replace(/\bp\.m\.?/g, "pm");
|
||||
if (/\bmorning\b|mañana|manhã|上午|صباح|maten/i.test(t)) return "09:00";
|
||||
if (/\bafternoon\b|tarde|après-midi|下午|مساء|aprèmidi/i.test(t)) return "13:00";
|
||||
const ampm = t.match(/\b(\d{1,2})(?::(\d{2}))?\s*(am|pm)\b/);
|
||||
const ampm = t.match(/\b(\d{1,2}):([0-5]\d)\s*(am|pm)\b/);
|
||||
const glued = !ampm ? t.match(/\b(\d{1,4})\s*(am|pm)\b/) : null;
|
||||
const clock = t.match(/\b([01]?\d|2[0-3]):([0-5]\d)\b/);
|
||||
if (ampm) {
|
||||
let h = parseInt(ampm[1]!);
|
||||
const m = ampm[2] ? parseInt(ampm[2]) : 0;
|
||||
const m = parseInt(ampm[2]!);
|
||||
if (ampm[3] === "pm" && h < 12) h += 12;
|
||||
if (ampm[3] === "am" && h === 12) h = 0;
|
||||
return `${String(h).padStart(2,"0")}:${String(m).padStart(2,"0")}`;
|
||||
}
|
||||
if (glued) {
|
||||
const digits = glued[1]!;
|
||||
let h: number, m: number;
|
||||
if (digits.length <= 2) { h = parseInt(digits); m = 0; }
|
||||
else { h = parseInt(digits.slice(0, -2)); m = parseInt(digits.slice(-2)); }
|
||||
if (h >= 1 && h <= 12 && m <= 59) {
|
||||
if (glued[2] === "pm" && h < 12) h += 12;
|
||||
if (glued[2] === "am" && h === 12) h = 0;
|
||||
return `${String(h).padStart(2,"0")}:${String(m).padStart(2,"0")}`;
|
||||
}
|
||||
}
|
||||
if (clock) return clock[0]!;
|
||||
try {
|
||||
const llm = getLlm(provider, apiKey, model);
|
||||
|
||||
@@ -12,7 +12,7 @@ import {
|
||||
getOfficeHoursDisplay,
|
||||
timeLabel,
|
||||
} from "../ai/reschedule-graph";
|
||||
import { getLlm, resolveAiProvider } from "../ai/llm-factory";
|
||||
import { getLlm, resolveAiProvider, type AiProvider } from "../ai/llm-factory";
|
||||
import { soundsLikeGoodbye } from "../ai/voice-assistant";
|
||||
import { getPublicBaseUrl } from "../utils/publicUrl";
|
||||
import {
|
||||
@@ -116,6 +116,64 @@ const NEW_OR_RESCHEDULE_Q: Record<string, string> = {
|
||||
"Haitian Creole": "Jis pou konfime — èske ou ta renmen pran yon nouvo randevou, oswa èske ou ta renmen reprogramè randevou aktyèl ou?",
|
||||
};
|
||||
|
||||
// ── Stale-conversation handling (SMS) ─────────────────────────────────────────
|
||||
// Unlike a phone call, a long gap between texts doesn't necessarily mean the
|
||||
// patient walked away — replies are often just slow. So instead of hard
|
||||
// resetting on any idle gap, we only second-guess the stored stage once the
|
||||
// gap crosses STALE_SMS_THRESHOLD_MS, and even then we ask the LLM to look at
|
||||
// the recent history before deciding whether the new text is still answering
|
||||
// the last question or reads like a fresh, unrelated conversation.
|
||||
|
||||
const STALE_SMS_THRESHOLD_MS = 3 * 60 * 60 * 1000; // 3 hours
|
||||
|
||||
const STILL_HERE_MSGS: Record<string, string> = {
|
||||
English: "It's been a while since our last message. How can I help you today?",
|
||||
Spanish: "Ha pasado un tiempo desde nuestro último mensaje. ¿Cómo puedo ayudarle hoy?",
|
||||
Portuguese: "Já faz um tempo desde nossa última mensagem. Como posso ajudá-lo hoje?",
|
||||
Mandarin: "距离我们上次的消息已经有一段时间了。今天我能帮您什么？",
|
||||
Cantonese: "距離我們上次嘅消息已經有一段時間喇。今日我可以點幫您？",
|
||||
Arabic: "لقد مر بعض الوقت منذ آخر رسالة لنا. كيف يمكنني مساعدتك اليوم؟",
|
||||
"Haitian Creole": "Sa gen yon ti tan depi dènye mesaj nou. Kijan mwen ka ede ou jodi a?",
|
||||
};
|
||||
|
||||
/**
|
||||
* After a long idle gap, ask the LLM whether the patient's new text still
|
||||
* answers what was last asked (recent history + stage in context) or reads
|
||||
* like an unrelated, fresh message. Defaults to "continue" on any failure —
|
||||
* an LLM call error shouldn't be the reason a legitimate reply gets bounced.
|
||||
*/
|
||||
async function isLikelyContinuation(
|
||||
recentHistory: { role: "user" | "assistant"; text: string }[],
|
||||
newMessage: string,
|
||||
stage: ConversationStage,
|
||||
apiKey: string,
|
||||
provider: AiProvider,
|
||||
model?: string,
|
||||
): Promise<boolean> {
|
||||
if (recentHistory.length === 0) return true;
|
||||
try {
|
||||
const llm = getLlm(provider, apiKey, model);
|
||||
const transcript = recentHistory
|
||||
.map(m => `${m.role === "assistant" ? "Office" : "Patient"}: ${m.text}`)
|
||||
.join("\n");
|
||||
const res = await llm.invoke([
|
||||
{
|
||||
role: "system",
|
||||
content:
|
||||
`You are helping a dental office text-messaging assistant resume a conversation with a patient after a long gap of silence. ` +
|
||||
`Here is the recent conversation (oldest first):\n${transcript}\n\n` +
|
||||
`The assistant is currently waiting on stage "${stage}" of a booking flow. ` +
|
||||
`The patient just sent a new message after the long gap: "${newMessage}"\n\n` +
|
||||
`Does this new message still look like it's answering/continuing that same conversation, or does it read like an unrelated, fresh message (e.g. a new question, a greeting, a different topic)? ` +
|
||||
`Reply with ONLY one word: "continue" or "restart".`,
|
||||
},
|
||||
]);
|
||||
return !/restart/i.test(String(res.content));
|
||||
} catch {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
|
||||
// ── POST /api/twilio/webhook/sms ──────────────────────────────────────────────
|
||||
|
||||
const CONVO_TIMEOUT_MS = 5 * 60 * 1000; // 5 minutes
|
||||
@@ -427,16 +485,33 @@ router.post("/webhook/sms", async (req: Request, res: Response): Promise<any> =>
|
||||
let stage = await getStage(patient.userId, patient.id);
|
||||
const chatTemplates = await storage.getAiChatTemplates(patient.userId);
|
||||
|
||||
// Reset conversation if idle for more than 5 minutes
|
||||
// Long idle gap: don't blindly resume (SMS replies are often just slow),
|
||||
// but don't blindly reset either — check whether the new text still reads
|
||||
// like a continuation before deciding.
|
||||
if (stage !== "initial" && stage !== "done") {
|
||||
const convRow = await db.patientConversation.findUnique({
|
||||
where: { patientId: patient.id },
|
||||
select: { updatedAt: true },
|
||||
});
|
||||
if (convRow?.updatedAt && Date.now() - convRow.updatedAt.getTime() > CONVO_TIMEOUT_MS) {
|
||||
if (convRow?.updatedAt && Date.now() - convRow.updatedAt.getTime() > STALE_SMS_THRESHOLD_MS) {
|
||||
const recentComms = await storage.getCommunicationsByPatient(patient.id);
|
||||
const recentHistory = recentComms.slice(-5).map((c: any) => ({
|
||||
role: (c.direction === "outbound" ? "assistant" : "user") as "assistant" | "user",
|
||||
text: c.body || "",
|
||||
}));
|
||||
const continuing = await isLikelyContinuation(
|
||||
recentHistory, Body, stage, activeAi.key, activeAi.provider, activeAi.model
|
||||
);
|
||||
if (!continuing) {
|
||||
clearPendingReschedule(patient.userId, patient.id);
|
||||
await setStage(patient.userId, patient.id, "initial");
|
||||
stage = "initial";
|
||||
const stillHereText = STILL_HERE_MSGS[language] ?? STILL_HERE_MSGS["English"]!;
|
||||
await saveOutbound(patient.id, stillHereText);
|
||||
res.set("Content-Type", "text/xml");
|
||||
return res.send(twimlMessages(stillHereText));
|
||||
}
|
||||
// Looks like a continuation — fall through and resume the stored stage.
|
||||
}
|
||||
}
|
||||
const officeContact = await storage.getOfficeContact(patient.userId);
|
||||
@@ -690,7 +765,7 @@ function aiVoiceTwiml(sayText: string, gatherActionUrl: string, locale: VoiceLoc
|
||||
// silence timeout — nothing after </Gather> would ever run.
|
||||
return `<?xml version="1.0" encoding="UTF-8"?>
|
||||
<Response>
|
||||
<Gather input="speech" language="${locale.gatherLang}" speechTimeout="auto" action="${gatherActionUrl}" method="POST">
|
||||
<Gather input="speech" language="${locale.gatherLang}" speechTimeout="auto" timeout="15" action="${gatherActionUrl}" method="POST">
|
||||
<Say voice="${locale.sayVoice}">${escapeXml(sayText)}</Say>
|
||||
</Gather>
|
||||
</Response>`;
|
||||
@@ -741,10 +816,16 @@ router.post("/webhook/ai-voice", async (req: Request, res: Response): Promise<an
|
||||
// A "recall" call (staff-triggered from Patient Connection, offering an
|
||||
// existing patient a routine exam & cleaning) seeds a different stage
|
||||
// than the default reminder-style check-in.
|
||||
const currentStage = await getStage(patient.userId, patient.id);
|
||||
if (currentStage === "initial" || currentStage === "done") {
|
||||
if (isRecallCall) await startRecallConversation(patient.userId, patient.id);
|
||||
else await resetConversation(patient.userId, patient.id);
|
||||
// Either kind of call always speaks its own pitch fresh (see greeting
|
||||
// below), so it must always reset the stage — even if a prior call of
|
||||
// either kind left the patient stuck mid-flow (e.g. "recall_asked_datetime"
|
||||
// or "asked_reschedule_datetime"). Otherwise the reply below gets
|
||||
// interpreted against that stale leftover stage instead of the fresh
|
||||
// pitch that was just spoken.
|
||||
if (isRecallCall) {
|
||||
await startRecallConversation(patient.userId, patient.id);
|
||||
} else {
|
||||
await resetConversation(patient.userId, patient.id);
|
||||
}
|
||||
|
||||
const officeContact = await storage.getOfficeContact(patient.userId);
|
||||
@@ -779,8 +860,12 @@ router.post("/webhook/ai-voice", async (req: Request, res: Response): Promise<an
|
||||
const locale = voiceLocaleFor(session.language);
|
||||
|
||||
// ── Caller stayed silent past the speech timeout ──────────────────────
|
||||
// A call is a single continuous conversation, unlike SMS — if it ends
|
||||
// here, mid-flow or not, there's nothing left to resume, so reset the
|
||||
// stage now instead of leaving the patient stuck for their next call/text.
|
||||
if (!SpeechResult?.trim()) {
|
||||
aiCallSessions.delete(CallSid);
|
||||
await setStage(session.userId, session.patientId, "done");
|
||||
return res.send(aiVoiceHangup("I didn't catch a response. Our staff will follow up with you if needed. Goodbye.", locale));
|
||||
}
|
||||
|
||||
@@ -795,6 +880,7 @@ router.post("/webhook/ai-voice", async (req: Request, res: Response): Promise<an
|
||||
// ── Caller said something that sounds like goodbye, or we've hit the cap ──
|
||||
if (soundsLikeGoodbye(SpeechResult) || session.turns >= MAX_AI_CALL_TURNS) {
|
||||
aiCallSessions.delete(CallSid);
|
||||
await setStage(session.userId, session.patientId, "done");
|
||||
return res.send(aiVoiceHangup("Thank you for calling, have a great day. Goodbye.", locale));
|
||||
}
|
||||
|
||||
@@ -802,6 +888,7 @@ router.post("/webhook/ai-voice", async (req: Request, res: Response): Promise<an
|
||||
const activeAi = resolveAiProvider(aiSettings ?? {});
|
||||
if (!activeAi) {
|
||||
aiCallSessions.delete(CallSid);
|
||||
await setStage(session.userId, session.patientId, "done");
|
||||
return res.send(aiVoiceHangup("Our AI assistant is not available right now. Our staff will follow up with you. Goodbye.", locale));
|
||||
}
|
||||
|
||||
@@ -826,6 +913,12 @@ router.post("/webhook/ai-voice", async (req: Request, res: Response): Promise<an
|
||||
|
||||
if (routed.nextStage === "done" || soundsLikeGoodbye(reply)) {
|
||||
aiCallSessions.delete(CallSid);
|
||||
// routePatientTurn() already persisted nextStage — but if we're hanging
|
||||
// up because the reply sounded like a goodbye while nextStage was still
|
||||
// mid-flow, force it to "done" so the next call/text doesn't resume there.
|
||||
if (routed.nextStage !== "done") {
|
||||
await setStage(session.userId, session.patientId, "done");
|
||||
}
|
||||
return res.send(aiVoiceHangup(reply, locale));
|
||||
}
|
||||
|
||||
|
||||
@@ -46,9 +46,16 @@ export async function startWebDavServer(): Promise<void> {
|
||||
|
||||
const config = readRcloneConfig();
|
||||
|
||||
// Serve only the backups folder — never the app root, which contains
|
||||
// .env (DB/Twilio/AI credentials) and the full source tree.
|
||||
const backupsDir = path.join(APP_ROOT, "backups");
|
||||
if (!fs.existsSync(backupsDir)) {
|
||||
fs.mkdirSync(backupsDir, { recursive: true });
|
||||
}
|
||||
|
||||
const args = [
|
||||
"serve", "webdav",
|
||||
APP_ROOT,
|
||||
backupsDir,
|
||||
"--addr", `:${config.serverPort}`,
|
||||
"--user", RCLONE_USER,
|
||||
"--pass", RCLONE_PASS,
|
||||
|
||||
@@ -5,7 +5,7 @@ import { prisma as db } from "@repo/db/client";
|
||||
export const DEFAULT_GREETING =
|
||||
"Hi {firstName}, this is Lisa, the AI assistant at {officeName}. How can I help you today?";
|
||||
export const DEFAULT_RECALL_GREETING =
|
||||
"Hi {firstName}, this is Lisa from {officeName}. We wanted to check in — you're due for a routine exam and cleaning, which most insurance plans cover at no cost. Would you like to schedule one?";
|
||||
"Hi {firstName}, this is Lisa from {officeName}. We wanted to check in — you're due for a routine exam and cleaning. Would you like to make an appointment because your insurance covers it?";
|
||||
|
||||
export type TwilioSettingsData = {
|
||||
accountSid: string;
|
||||
|
||||
@@ -53,7 +53,7 @@ const CALL_DEFAULTS = {
|
||||
newPatientGreeting: "No problem, let's get you set up.",
|
||||
generalFallback: "Sure, how can I help?",
|
||||
recallGreeting:
|
||||
"Hi {firstName}, this is Lisa from {officeName}. We wanted to check in — you're due for a routine exam and cleaning, which most insurance plans cover at no cost. Would you like to schedule one?",
|
||||
"Hi {firstName}, this is Lisa from {officeName}. We wanted to check in — you're due for a routine exam and cleaning. Would you like to make an appointment because your insurance covers it?",
|
||||
};
|
||||
|
||||
const DEFAULT_SMS_TEMPLATES = [
|
||||
|
||||
Reference in New Issue
Block a user