Frequently Asked Questions

No. Your existing landline number can be forwarded to the internet-based system seamlessly.

As long as your business complies with local regulations and has SIP communication requirements, you can apply.

If your current PBX system supports SIP Trunk, no additional equipment is needed. If you are using a standard landline, a small VoIP Gateway device will be required to route the signal to the internet.

The AI itself has virtually unlimited call-handling capacity. The actual number of simultaneous calls depends on the number of concurrent lines allocated by your telecom provider.

Traditional voicemail can only record messages. Our AI engages in real-time conversations with callers, provides accurate answers, and generates a summary for you immediately after each call.

Not at all. We use Gemini Live technology, which produces natural-sounding speech — not pre-recorded audio. It sounds like a professional human receptionist.

The AI strictly responds based on your company's knowledge base. For issues it cannot resolve, it will immediately transfer the call to a live agent.

All data is stored in Google Cloud's regional data centers with AES-256 military-grade encryption, fully compliant with data protection regulations.

No. Google ensures enterprise data isolation. Your privacy is fully protected and never shared.

Yes. We provide API integration services — simply share your API documentation and we will assist with the setup.

We offer a Proof of Concept (PoC) program. If the solution does not meet the agreed standards during the trial phase, a refund can be arranged per the agreement.

The basic plan setup fee is RM7,000, which includes foundational configuration and notification integration.

The monthly RM500 maintenance fee covers Google Cloud resources, ongoing system optimization, and stability monitoring.

Each inbound call handled by the AI is charged at RM1.50 per call, billed at the end of each month.

The core processing runs on Google Kubernetes Engine, providing telecom-grade reliability and auto-scaling capabilities.

Yes. Gemini Live has powerful multilingual capabilities and can automatically detect the caller's language, supporting over 70 languages including English, Malay, Chinese, Japanese, Korean, and more.

No. AI only answers from the knowledge base you create—for example, filing deadlines, required documents, and fee explanations. It never provides tax judgments or filing advice. Questions outside scope are always transferred to the appropriate accountant.

Yes. After understanding the caller's intent, AI automatically routes to the corresponding department or person. Callers don't need to navigate a keypad menu—just state their need and they're correctly connected on the first try.

It can complete bookings directly. Through booking system integration, AI checks trial class slots in real time, confirms the child's grade and subject, completes the booking during the call, and auto-sends a confirmation SMS. Admin just needs to verify the next day.

Yes, through API integration. After answering, AI checks order status, shipment progress, and return/exchange info in real time. Customers provide an order number and get an answer within seconds.

No. AI only answers from the knowledge base you create—for example, clinic hours, appointment procedures, and department descriptions. It never provides medical diagnoses or medication advice. Questions outside scope are always transferred to human medical staff.

It can complete bookings directly. Through booking system integration, AI checks room availability in real time, confirms dates and room types, completes the reservation during the call, and auto-sends a confirmation SMS to the guest.

AI only answers from the knowledge base you've reviewed and approved. Content is set up and maintained by HR. Questions outside the knowledge base are always transferred to a human HR staff member—AI never speculates or provides unverified information.

No. When AI detects it can't answer or the customer requests a human, it seamlessly transfers to a human agent with the call summary included. The customer doesn't need to repeat themselves—the experience flows uninterrupted.

Yes. Through policy system integration, policyholders verify their identity and AI instantly pulls up coverage details and claims progress, answering during the call—no human lookup needed. Policyholder experience dramatically improves.

Yes, through API integration with major logistics management platforms. After answering, AI checks package location, estimated delivery time, and exception status in real time, responding to recipients within seconds—no manual lookup needed.

It can complete reservations directly. Through booking system integration, AI checks availability in real time, confirms time slots and party size, completes the booking during the call, and auto-sends a confirmation SMS to the guest.

Yes. After connecting to CRM, AI matches caller numbers to auto-pull client records. When reps take over, they already have the full background—no need for clients to repeat themselves. Every call feels like they've known each other for years.

Yes. The knowledge base is built by you. Once you configure property data, common terms, and standard Q&A, AI answers only from that content, ensuring every response fits your business context.

Common automated actions include: writing structured data to CRM, pushing call summaries to Google Chat or Slack, auto-creating booking or task entries, sending confirmation SMS, and triggering the next outbound call task. Actions can be flexibly configured based on your business needs.

In the Gemini AI agent's system prompt, you pre-define what information to extract and what conditions trigger which actions. For example, 'if the customer mentions wanting to schedule an appointment, extract the preferred date and time and auto-create a booking entry.' The AI executes these rules automatically.

Action execution rules can be set with confidence thresholds. Low-confidence results get pushed to human review via Google Chat rather than auto-executing. The prompt can also be optimized based on actual data to progressively improve accuracy.

Data is stored in your own GCP project—BigQuery, Firestore, Cloud SQL, and other options. You retain full data ownership and can use any analytical tool for subsequent analysis.

Currently supports integration with Google Calendar, Cal.com, and other booking systems with API access. For other systems, custom API integration is also possible—please contact us for details.

The AI agent queries the connected booking system via API in real time to get available time slots. The customer hears available options during the call, and the response time is nearly instant with no noticeable delay.

Cancellation or rescheduling operations can also be set up through the AI agent. When the customer calls back, the AI identifies the existing booking, processes the requested changes, and auto-updates the booking system.

Fully automatic. Once a booking is confirmed, the system auto-sends a confirmation SMS to the customer, syncs the record to the booking system, and notifies staff via Google Chat—the entire process requires no manual operation.

The system monitors three conditions at every turn: ① customer's question falls outside the AI knowledge base scope, ② customer explicitly requests a human agent, ③ emotional intensity exceeds the set threshold. Any condition met triggers the transfer process.

A Google Chat message is sent containing the conversation summary, customer's main intent, emotional state, and relevant background information, so the agent understands the situation before picking up.

Yes. Transfer rules can be configured by department, specific person, or even by another AI agent. For example, appointment-related calls go to Booking AI, pricing inquiries to the sales team, and complaints to senior representatives.

The customer hears a brief hold prompt (e.g., 'One moment please, I'm connecting you to a specialist'). Actual transfer times depend on your phone system's SIP Trunk settings, typically completing within a few seconds.

Yes. Both can be enabled at the same time. The Gemini AI agent checks them in sequence when a call comes in—allowlisted numbers get direct transfer, blocklisted numbers are auto-blocked, and all other calls proceed to the standard AI handling flow.

You can add them manually via the admin interface, or sync them via API with your CRM so existing customer data automatically becomes allowlist entries—no duplicate entry needed.

When a blocklisted number calls, the Gemini AI agent auto-rejects or plays a preset voice message. The message content can be customized. The caller won't know they've been blocklisted.

Withheld numbers can't be matched against allowlist or blocklist, so they default to the standard AI handling flow. If you need special handling for withheld numbers, this can be configured separately in the settings.

Recording files are stored in your own Google Cloud Storage, within your Google Cloud project. We do not retain any recording data—you have complete data ownership.

Call logs are automatically written to BigQuery in your GCP project. You can use any BigQuery-compatible tool for analysis, such as Looker Studio, Tableau, Python, etc. We do not provide dashboard creation services.

Retention period is set by you. Since data is stored in your Google Cloud Storage, you can freely adjust retention duration and access permissions according to company policies or regulatory requirements.

Recording data is stored in your own GCP project—we never access or obtain any personal data. Specific compliance configurations (encryption, access control, etc.) should be set up according to your security policies on the GCP side.

After a call ends, the transcript is auto-written to BigQuery. Gemini then performs semantic analysis on the content, auto-determining categories and applying tags based on preset classification rules. Classification rules can be customized according to your business needs.

Tag content can be freely configured according to your needs. Common examples include: issue type (new customer inquiry, complaint, cancellation discussion), opportunity grade (high potential, under evaluation), and risk flags (urgent processing, compliance anomaly). Initial setup is assisted by our team.

Based on tag content, the system can auto-execute follow-up workflows—for example: 'Complaint' tags auto-send Chat notifications to managers, 'High potential' tags auto-create CRM opportunity tasks, and 'Callback needed' tags auto-assign to the corresponding sales representative.

Call classification data is stored in BigQuery within your own GCP project. You can freely analyze it using any BigQuery-compatible tool, such as Looker Studio, Tableau, Python, etc. We do not provide dashboard creation services—tool selection and usage are up to you.

As the call connects, the system auto-retrieves the caller's number and matches it against the CRM database. If a matching record is found, AI greets the customer by name or company name from the very first sentence. The entire process completes before the conversation begins—the customer notices no delay.

The Gemini AI agent auto-requests the caller's name and contact information, records it to CRM as the basis for a new customer profile. Follow-up callbacks or SMS notifications can then be scheduled.

Landlines can't receive SMS, so when the system detects a landline call, the Gemini AI agent auto-requests the caller's mobile number, records it to CRM, and uses it for subsequent SMS delivery.

Yes. For example, known VIP customers can be directly transferred to their dedicated sales rep, regular known customers enter the standard service flow, and unknown callers first have their info collected before a callback is scheduled—three separate flows can be configured independently.

Currently supports Google Chat, Slack, and LINE. Notifications are sent via each platform's API to designated channels or groups. Integration requires you to provide the corresponding platform's API key or Webhook URL.

Yes. Based on IVR routing or conversation analysis results, notifications can be directed to corresponding channels—for example, sales inquiries notify the sales team, complaints notify the customer service manager. Each department only receives notifications relevant to them.

Notification content includes: caller number, customer name (if CRM has a record), call time, Gemini-generated conversation summary, and determination results (e.g., callback needed, complaint, FAQ miss, etc.).

When the Gemini AI agent encounters a question not in the knowledge base, it auto-sends the question content to the designated channel. Accumulating these missed questions serves as the basis for improving FAQs and the knowledge base, continuously improving answer quality.

Function Calling is a Gemini capability that allows AI to automatically invoke predefined data structures when specific conditions are met during conversation, converting dialogue content into structured data. Designers only need to define 'which fields are needed, which are required'—Gemini automatically collects and fills in corresponding values during natural conversation.

Yes. Required fields (e.g., name · phone · preferred date) and optional fields (e.g., notes · preferred staff) can be freely configured based on business needs. Gemini ensures all required fields are collected before ending the collection process; optional fields are not forcefully asked if the customer doesn't mention them.

Format validation rules can be set in field definitions—for example, 'phone number must be a valid format.' If the customer's answer doesn't match, Gemini naturally asks them to clarify—for example, 'Could you please tell me your phone number?'—rather than displaying an error message.

For customers, one phone call handles everything—no need to open a webpage, type text, or submit a form. For businesses, data is auto-organized by AI into structured format, saving manual data entry time and avoiding input errors. Especially suitable for elderly customers or non-digital-native users.

Gemini understands conversation context. If a customer says 'Actually, change that date to the day after tomorrow,' Gemini auto-updates the corresponding field value without needing to re-confirm from the beginning.

Traditional IVR only allows selecting fixed options via keypad, with pre-set branch paths that can't be flexibly adjusted—customers often feel frustrated when options don't match their needs. AI IVR supports natural language input, and the Gemini AI agent can understand phrases like 'I want to reschedule my appointment,' dynamically determining the routing target without requiring the customer to memorize menu numbers.

Dividing AI agents by business role (Receptionist AI, Booking AI, FAQ AI, Billing AI, Complaint AI, etc.) is key to building a high-quality AI phone system. Each AI agent focuses only on its own business scope, allowing fine-tuned knowledge base adjustments for that domain, avoiding the problem of a single AI trying to do everything and doing nothing well. The Receptionist AI handles intent understanding and routing; specialized AIs handle deep processing. Clear division of labor maximizes accuracy at every stage and makes future maintenance and expansion easier.

Multiple fallback mechanisms are in place: first, the Gemini AI agent asks the customer to repeat or rephrase; if still unable to understand, it auto-prompts 'You can also use the keypad—press 1 for appointments, press 2 for inquiries...' switching to DTMF mode; if still unresolved, it transfers to a human agent, ensuring no call goes unhandled.

Yes. For example, if a customer says 'I want to make an appointment' while talking to the FAQ AI, the system instantly switches the call to the Booking AI, continuing the conversation without interruption—the customer doesn't need to re-explain their needs.

Two analogies work well. First: 'guiding a guest to different rooms'—like walking into a hospital where you first tell the reception desk what you need, and they direct you to the right department where a specialist handles your specific issue. The Receptionist AI is the reception desk, specialized AI agents are the department specialists, each serving in their own space. Second: 'switching website pages'—like clicking from a homepage to a feature page on a website, where the backend loads completely different systems and data, but you feel like you're browsing the same site. AI agent switching works the same way—for the customer it's one continuous phone call, but behind the scenes, the system has quietly switched to the AI best suited to handle their current need.

Detection completes within an extremely short time after the conversation begins. The customer experiences virtually no delay, and the conversation feels indistinguishable from native-language service.

Mid-call switching is not supported. AI determines the language based on what the speaker uses at the start of the call and maintains that language throughout. If a different language service is needed, please call again.

Powered by Gemini Flash Live, it supports Japanese, Traditional Chinese, Simplified Chinese, Cantonese, English, and other major languages, with ongoing expansion. Please contact us for specific language support inquiries.

Yes. Gemini Flash Live has significantly enhanced noise resistance, enabling accurate speaker recognition even in outdoor or in-car noisy environments.

The Gemini AI agent uses high-quality voice synthesis technology with natural, fluid intonation. However, we recommend proactively informing customers at the start of the conversation with 'Hello, I'm an AI assistant'—this isn't just about integrity, it also helps set correct expectations for the service.

The Gemini AI agent has no simultaneous call limit. Regardless of how many calls come in at once, the system can answer them all and conduct independent natural language conversations without interference.

Yes. Through SIP Trunk connection, your existing landline or toll-free numbers can be directly routed to the Gemini AI agent—no number changes needed for customers.

Yes. Settings can be configured based on your business hours—for example, humans answer during business hours, then auto-switch to the Gemini AI agent after hours. Or different answering methods can be set by call type.

When the Gemini AI agent determines a question is beyond its scope, or the customer explicitly requests a human transfer, the system can auto-transfer to the designated human agent or department. Pre-transfer, call background is sent via Google Chat so the agent is fully prepared before answering.

Regular AI may speculate and generate answers based on training data, risking 'hallucination'—AI fabricating nonexistent information. RAG strictly limits AI to answering only from your provided data, ensuring every response has supporting evidence. This is especially suitable for business scenarios requiring precision, such as product specifications, medical information, and regulatory content.

Phone AI answers need to complete in extremely short timeframes, so we recommend prioritizing structured databases. Database queries are faster, helping reduce response latency and improve the call experience. If data is already in PDF or DOCX format, unstructured files work too, but we recommend simplifying document structure to improve retrieval accuracy.

Yes—your information needs to be organized and submitted by you. We assist with database setup and import, but the data content itself (product information, FAQs, business rules, etc.) needs to be prepared and managed by you.

After updating data, the Vertex AI Search index needs to be rebuilt. Once complete, AI can answer based on the latest content. Index update time depends on data volume, typically completing within minutes to tens of minutes.

SMS delivery is executed via API calls to an external SMS service—it's not a built-in feature. It can be connected to an SMS service provider suited to your needs; the delivery process is fully automated with no manual operation required.

Landlines can't directly receive SMS, so when the system determines the call is from a landline, the Gemini AI agent auto-asks the customer for their mobile number during the call, then proceeds with SMS delivery after obtaining it.

Yes. For example, in a booking flow, the Gemini AI agent first collects customer information then ends the call. After the backend booking system confirms, SMS auto-triggers to notify the customer of the booking result—the entire process requires no human intervention.

Any text content—for example, form URLs, Google Maps links, LINE official account links, email addresses, appointment confirmation numbers, etc. Information that's difficult to convey verbally over the phone can be precisely delivered via SMS.

No problem. AI is optimized for local language contexts and can accurately understand intent and record bookings regardless of dialect or accent.

If walk-ins, manual bookings, and AI all use the same real-time-updated booking system, AI will schedule based on the latest data to avoid double bookings. Actual effectiveness depends on your existing system's sync mechanisms.

If integrated with an SMS delivery system, AI can auto-send a confirmation SMS after the call, including booking time, location link, or confirmation code for easy reference.

Using Gemini AI's Natural Language Understanding (NLU) technology, the system makes real-time judgments based on the caller's actual conversational intent. Unlike traditional number-based blocking, AI first answers and politely asks about their needs—only filtering when confirmed as advertising or malicious calls. All calls have structured records, ensuring 100% of your important customer opportunities are preserved.

Yes. AI first answers the call and understands the caller's needs, then uses conversation content and company rules to help identify solicitation, advertising, or cold calls. For real estate, it reduces non-target prospecting interruptions; for restaurants, it decreases phone lines being occupied by sales calls, letting staff focus on in-person service.

Absolutely not. We use the highest-grade voice synthesis and recognition technology, achieving near-human response times. The AI agent supports multiple language recognition, understanding caller needs and providing appropriate responses—delivering a more professional, faster first impression than human transfers.

Results can be verified within 7 days of deployment. Specific benefits include: 1. Freed-up staff productivity: eliminating the hidden cost of employees handling sales calls. 2. Faster opportunity organization: call content auto-classified and transcribed for follow-up and management. 3. Improved communication quality: ensuring lines are always open when real customers call.

The system auto-redials based on your settings, or switches to SMS notification after multiple unanswered attempts, ensuring important reminders (like follow-up visits, payments) are actually delivered.

Yes. AI has bidirectional interaction capability, collecting customer intent directly during conversation and syncing results to your management backend in real time.

We use compliant business phone numbers for outbound calls, and AI conversations are highly natural and polite, significantly reducing the risk of rejection and improving reminder completion rates.

AI requires the recipient to provide voice confirmation during the call (e.g., saying 'received' or 'handling it'). If no confirmation is received, it continues dialing backup personnel on the list.

Yes, it can be connected to sensors or monitoring signals. Once an anomaly is detected, AI immediately initiates the emergency dispatch process without any manual operation. Actual integration success depends on API availability and other details.

We use RAG technology to read from your company manuals, product documentation, or website information, ensuring AI answers are evidence-based and always up-to-date.

No. When AI determines a question exceeds its knowledge scope, it honestly informs the caller and proactively guides them to human consultation, or records the question for an agent to call back.

Yes. AI can extract human voice from background noise, ensuring accurate data entry even in field sites, factories, and similar environments.

Nearly instant. Within seconds of the call ending, AI completes voice-to-text conversion and writes the notification to your chat tools.

AI analyzes intent signals in the conversation (e.g., price inquiry, booking a demo, requesting a quote) along with the customer's emotion and tone, comprehensively assessing deal potential.

AI auto-tags high-potential customers as 'hot opportunities' and pushes real-time notifications to sales via LINE, Slack, or the CRM system.

Built with Gemini, AI conducts natural conversations without sounding robotic.

Yes. AI has instant language detection capability—once it detects the customer has switched languages, AI immediately follows suit, ensuring the conversation continues uninterrupted.

AI performs identity verification before any lookup (e.g., confirming last digits of phone number or order number), ensuring only the customer themselves can access the relevant information.

As long as your database permissions allow it, AI can verify the information during the call and directly update the backend address, simultaneously sending an SMS notification of the change.

When the database has issues or no matching order number is found, AI proactively transfers the call to a human agent—preventing customer frustration with the automated system.

Built with Gemini, conversations are fluent with natural emotional variation, effectively reducing the rejection that cold calls typically trigger.

AI auto-identifies voicemail, disconnected numbers, and instant hang-ups, only passing prospects who show product interest and willingness to engage to sales for follow-up.

Yes. Once your price list or system database is connected to AI's knowledge base, AI reads the latest data directly—no pricing inconsistencies.

AI's answer sources can be configured to your company's product, pricing, or quote data. As long as data is properly maintained and synced in real time, AI responds based on the latest information. Actual content should always be verified against your company's final data.

AI evaluates based on preset standard Q&A, analyzing the professionalism and completeness of responses, then produces an assessment report with grading and recommendations.

AI's role is 'efficient pre-screening.' It helps HR handle initial verification and assessment for high-volume entry-level positions—HR only needs to spend time interviewing the candidates who pass the initial screen.

Recognition is based on the phone numbers in your CRM database. As long as a caller is tagged as VIP in your system, AI activates the dedicated flow the instant the phone rings.

If connected to CRM or customer data systems, AI can perform priority transfers based on preset rules while simultaneously sending customer information to the designated platform, helping the agent quickly understand the call's context.