Tilicho Labs⌘KTalk to us
Online retail, marketplaces and travel bookings

Voice agents for e-commerce and travel in Southeast Asia

AI agents that confirm cash-on-delivery orders, call when a delivery fails, answer your helpline about orders and bookings, and call travellers when a booking needs attention.

Cash-on-delivery order before dispatchBahasa Indonesia
Sample call
AgentCustomerSystems
Order read: sneakers, size 42, Rp459.000, cash on delivery, placed 2 hours ago. First order from this number
AgentSelamat siang, apakah saya berbicara dengan Bapak Andi Pratama?
CustomerIya, betul.
Customer confirmed
AgentSaya asisten AI dari ABC Store, dan panggilan ini direkam. Bapak memesan sneakers ukuran 42, Rp459.000, bayar di tempat. Apakah pesanan ini saya konfirmasi?
CustomerIya, konfirmasi. Tapi nomor rumahnya salah, harusnya 17, bukan 71.
Address corrected: No. 17. Landmark added: next to the mosque
AgentSudah saya ubah ke nomor 17. Kalau Bapak bayar sekarang lewat transfer atau e-wallet, tidak perlu siapkan uang tunai. Mau saya kirim link-nya lewat WhatsApp?
CustomerNggak usah, bayar di tempat saja.
AgentBaik, Pak. Pesanan dikirim besok dan sampai paling lambat hari Kamis. Detailnya saya kirim lewat WhatsApp.
Order confirmed and released for dispatch
WhatsApp sent: order summary, delivery date, tracking link
Courier update: failed delivery, address not found. Order: rice cooker, ₱2,490, cash on delivery. Second attempt allowed
AgentHello po, si Maria Santos po ba ito?
CustomerOo, ako nga.
Customer confirmed
AgentIto po ang AI assistant ng ABC Store, at recorded ang call na ito. Hindi po nahanap ng rider ang address ninyo kahapon para sa rice cooker. Pwede po ba kayong magbigay ng landmark?
CustomerAy sorry. Sa tabi ng sari-sari store sa Mabini Street, yung green gate.
AgentNoted po. Kailan po kayo nasa bahay para sa delivery?
CustomerBukas ng umaga.
Landmark added: beside the sari-sari store on Mabini Street, green gate
Reattempt booked: tomorrow, 9 am to 12 pm. Update sent to the courier
AgentOkay po, bukas ng umaga. Pakihanda lang po ang ₱2,490 cash. Magse-send ako ng SMS with the tracking link.
SMS sent: delivery date and tracking link
Caller number matched: baju kurung, size M, RM129 prepaid, delivered Monday. Inside the 7-day return window
AgentTerima kasih kerana menghubungi ABC Store. Saya AI assistant, dan panggilan ini direkodkan. Puan call pasal baju kurung yang sampai hari Isnin ke?
CustomerYa. Saiz M tu kecil sangat, I nak return.
AgentMasih dalam tempoh return. Saya boleh tukar ke saiz lain, atau buat refund. Puan nak yang mana?
CustomerAda saiz L tak?
Stock checked: size L available
AgentSaiz L ada stok. Courier akan ambil yang M dan hantar yang L sekali, hari Jumaat. Boleh saya book?
CustomerOkay, boleh.
Exchange raised: M to L. Pickup and delivery booked: Fri
WhatsApp sent: exchange number, date, packing instructions
Booking read: Bangkok to Chiang Mai, 22 Nov, 2 adults. Airline moved the flight from 7:10 am to 3:40 pm. Hotel check-in 2 pm
Agentสวัสดีค่ะ ขอเรียนสายคุณนภา ศรีสุข ค่ะ
Customerค่ะ พูดอยู่ค่ะ
Lead traveller confirmed
Agentดิฉันเป็นผู้ช่วย AI จาก ABC Travel ค่ะ การสนทนานี้มีการบันทึกเสียง สายการบินได้เปลี่ยนเวลาเที่ยวบินกรุงเทพฯ ไปเชียงใหม่ของคุณ วันที่ 22 พฤศจิกายน จาก 7.10 น. เป็น 15.40 น. ค่ะ
Customerแบบนี้เสียไปครึ่งวันเลย มีเที่ยวบินอื่นไหมคะ
Alternatives returned: same airline 9:25 am, no fare difference; another airline 6:05 am, ฿850 more per person
Agentมีเที่ยวบิน 9.25 น. ของสายการบินเดิม ไม่มีค่าใช้จ่ายเพิ่มค่ะ หรือเที่ยวบิน 6.05 น. เพิ่มคนละ 850 บาท หรือถ้าต้องการขอคืนเงินก็ได้ค่ะ
Customerเอาเที่ยว 9.25 น. ค่ะ
Rebooked: 22 Nov, 9:25 am, 2 adults, no fare difference
SMS sent: new e-ticket and itinerary
One record acrossVoiceWhatsAppEmail
Names, amounts and dates are examples.
Minutesfrom a COD order to the confirmation call

Today: orders wait in a queue on a sale day

50,000+calls a day across the helpline and campaigns

Today: about 100 per person

25+languages, switching during the call

Today: the languages your team speaks

Every callrecorded, transcribed and tagged with the outcome

Today: a sample of calls is reviewed

30+voice agents in production
In production in
LendingReal estateHigher educationHealthcareE-commerceHome services
Why Tilicho

What is different about how we build it

Not tied to one model

Companies that sell their own voice model run every call on that model. We choose the model for each step, such as Gemini 3.8 Live for the conversation and Chirp 3 for speech recognition, and change it when a better one is released.

One conversation across voice, WhatsApp and email

A caller who hangs up and replies on WhatsApp, or writes in by email, reaches the same agent with the same history. Every channel reads and writes the same customer record.

Sneakers · Rp459.000 · cash on delivery
  1. Voice · Day 1COD confirmation call. Address corrected, order released for dispatch
  2. WhatsApp · Day 1Order summary and tracking link. Customer asks about the size, answered
  3. Voice · Day 4Courier marks the delivery failed. Agent calls and books Saturday
  4. Email · Day 6Delivery confirmation and invoice

A team that sets it up and stays

A small Tilicho team writes the call flows with your support and operations teams, connects your systems, listens to calls every week and makes the changes.

First use case live in a month

Accelerators for 25 use cases come with the call flow, the opening, the system reads and writes and the measures already built.

Runs in your cloud

The agent runs in your own Google Cloud project. Recordings, transcripts and customer data stay with you.

Where to start

Three agents on one customer record

Orders, the helpline and travel bookings, each with its own agent. All three read and write the same customer, order and booking record, so one integration supports all three.

The order or booking is confirmedIt is delivered, or the trip goes aheadThe customer reviews it, and orders or books again
Use cases

25 accelerators for e-commerce and travel

Each agent starts from an accelerator, so a use case is on live calls in about one month.

Orders and delivery

Travel bookings

The detail

For the teams who will run it

Operations, compliance and technology. Open what you need.

Support team and voice agentRow by row, today and with the agent
Support teamVoice agent
Calls in a dayAbout 100 per person50,000+ across the helpline and campaigns
On a sale dayCOD orders wait in a queue to be confirmedOrders called as they come in
LanguagesThe languages your team speaks, often one desk per country25+, switching during the call
Order or bookingLooked up by order number while the customer waitsRead from the record before the first question
ScriptVaries by person and by dayThe approved script on every call
RecordA sample of calls is reviewedRecording, transcript and outcome tag on every call
CapacityHiring and training ahead of each sale seasonAdded in a day, removed after the peak
  • One Tilicho e-commerce deployment on WhatsApp handles 70% of calls, chats and emails without a person and recovers about 4% of abandoned carts.
What the agent can doIn the conversation, during the call, after it and with your people

In conversation

  • Answers and places calls, 50,000+ in a day
  • Speaks 25+ languages and switches when the customer does
  • Stops when interrupted and picks up from there
  • Stays clear on a busy street, in a market or at a station
  • Takes order numbers, booking codes, postcodes and addresses read aloud
  • Remembers earlier calls and chats with the same customer

Actions during the call

  • Reads and updates the customer, order and booking record
  • Confirms COD orders and corrects addresses before dispatch
  • Books return pickups and new delivery dates through your shipping system
  • Rebooks travel from the options your booking system returns
  • Sends payment links, tracking links and e-tickets on WhatsApp or SMS
  • Answers from your return policy, fare rules and FAQs

Outbound campaigns

  • Campaigns from your order system, booking system or CRM
  • Targeting by country, order status, payment method, travel date or last purchase
  • Retries for no-answers and scheduled callbacks
  • Follow-ups until the outcome is reached
  • Answering machines and voicemail detected

After every call

  • Recording and full transcript
  • Summary with the key points and next actions
  • Customer sentiment
  • Outcome tag written to your system
  • Summary message to the customer and to your team

With your people

  • Live transfer with the order or booking attached
  • Fraud and stranded-traveller cases go to a person at once
  • Approval from a supervisor on Slack, such as a goodwill refund
  • Callback booked for a named team member

Connected to your stack

  • Salesforce, HubSpot, Zoho, LeadSquared, KaptureCX
  • Twilio, Plivo, Telnyx, Exotel, Genesys, SIP trunks
  • Zendesk, Freshdesk, Shopify, Google Calendar, Calendly
  • 5,000+ more apps through Zapier
  • Shipping systems, couriers and booking engines through their API
Before, during and after the callWhat the agent reads, does and writes back
1

Before the ring

The number is matched to the customer's orders and bookings: items, payment method, courier status, travel dates and open tickets. Calling hours and consent are checked before an outbound dial.

2

During the call

The agent confirms the order or the booking, takes the change the customer asks for, and corrects the address, books the pickup, rebooks the trip or sends the payment link while the customer is on the line.

3

When a person is needed

A fraud concern, a refund dispute, a traveller stuck at an airport or station, or any request for a person goes to your team live, with the order or booking, the history and the summary on screen.

4

After hang-up

Status, outcome tag, transcript and recording are written to your order or booking system and CRM. The courier update, the payment follow-up or the review call is scheduled.

Live in one monthWhat is ready in the accelerator, what we do with you, and the timeline
Ready in the acceleratorDay 1
  • The call flow and prompts for the use case
  • The fixed opening: who is calling, why, AI and recording disclosure
  • Outcome tags and the number it is measured by
  • Integrations for telephony, WhatsApp, SMS, Shopify, CRM and ticketing
  • A test set of calls for the common situations
Done with youWeeks 1 to 4
  • Your numbers and telephony connected in each country
  • APIs to your order, shipping, booking and payment systems
  • Your approved script, return policy, fare rules and languages
  • Information-security sign-off
  • Staging tests on your data, then the first live calls
Live in one month
2 weeks
when your store is on standard integrations, such as Shopify, and the script is already approved
1 month
the usual case: one or two system APIs, such as the shipping system and the booking engine
2 months
when a system such as an ERP needs a new API or the security review runs long
How every call opensThe fixed opening, the calling window and the controls
1

Right person

On an outbound call the agent confirms it is speaking to the customer who placed the order, or the lead traveller, before it names the order or the trip.

2

Who is calling and why

Your brand's name and the purpose of the call, in the words your team approved.

3

AI assistant, recorded call

The agent says it is an AI assistant and that the call is recorded.

4

The conversation

Only then does the script move to the order, the delivery or the booking.

Calling window
12 am9 am9 pm12 am

Calling hours

Offer calls, such as cart recovery and repeat-purchase offers, are placed only inside the hours you set, for example 9 am to 9 pm in the customer's time zone. The helpline answers at any hour.

Consent and Do Not Call

Offer calls go only to customers who have consented. For Singapore numbers, the Do Not Call Registry is checked before a marketing call.

Payments by link

The agent never takes a card number, OTP or PIN on the call. It sends a secure payment link from your payment gateway.

Personal data

PII is masked in transcripts and logs. Access is by role. Consent, retention and access are set for each market's personal-data law: PDPA in Singapore, PDPA 2010 in Malaysia, UU PDP 2022 in Indonesia, PDPA in Thailand, the Data Privacy Act of 2012 in the Philippines, the Personal Data Protection Law of 2025 in Vietnam.

Your own cloud project

The agent runs in your Google Cloud project, in the region you choose, such as Singapore or Jakarta. Recordings and customer data stay in your storage.

Changes to an order

The agent changes an address, an order or a booking only for the confirmed customer, and reads the change back before it is saved.

How the agent is builtPipelines, turn-taking and what keeps a call reliable
Speech to speechOne model hears and speaks. The fastest replies, and the caller's tone carries through. The default for live calls.
Caller audioGemini 3.8 LiveAgent audio
CascadedThree models in sequence. Each step can be inspected, and the model cost per minute is lower. Used where the script is strict.
Caller audioChirp 3 speech to textGemini 3.8 FlashGemini 3.8 Flash TTSAgent audio

One turn of a call

CallerAgent Speech detected (VAD)A pause, not the endEnd of turnAgent repliesCaller interruptsAgent stopsAgent answers the new point
Voice activity detection (VAD)Separates the caller's speech from line noise and silence. Only speech is sent to the model, so silence is neither processed nor paid for.
End-of-turn detectionTells a pause in the middle of a sentence from the end of it, so the agent does not reply while the customer is still reading out an address or a booking code.
Barge-inWhen the caller interrupts, the agent stops speaking at once and the words it did not say are dropped from the conversation.
Answering machine detectionDecided from the audio in the first seconds, before a language model is involved, so voicemail is not counted as a contact.
Noise handlingA motorbike in the street, a wet market, rain on a tin roof, an airport gate. The agent stays clear without asking the customer to repeat.
Context compressionA live model re-reads the call so far on every turn. Older turns are summarised, so reply time and cost stay flat on long calls.

Reliability

In-region deployment

The agent runs in the Google Cloud region closest to your customers, for example Jakarta or Singapore. This is fixed before the build, since reply time cannot be tuned afterwards.

Language mirroring

The agent mirrors the customer turn by turn, so it follows the customer's language and not a setting.

Supervisor and task agents

A supervisor routes to small agents that share the state of the call. The script grows by adding agents.

Supervisor
Order lookupBookingPayment

Model fallback

A health-checked fallback model in the same region takes over mid-call, with the state of the call intact.

Warm transfer

Your team member sees the order or booking, the history, the transcript summary and the captured details before speaking.

Closing the loop

An analyzer agent keeps each order or booking open until it reaches an end state: confirmed, delivered, rebooked, refunded, or handed to your team.

Where it sitsYour systems, the agent, the cloud project and the models
Your systemsOwned by you
Orders and bookingsCarts, orders, returns and refunds, or bookings, booking codes and itineraries. Shopify, your own platform or your booking engine. Two-way sync through APIs.
Shipping and suppliersYour shipping system and courier partners in each country, and the airline, rail, bus, ferry and hotel feeds behind your bookings.
Telephony and paymentsTwilio, Genesys or SIP trunking on your existing numbers in each country. Your payment gateway links.
Voice layerBuilt and run by Tilicho
Call handlingCall transport, latency control, live call state, prompts and tool calls, PII redaction.
Speech modelsKnows when the caller has finished, hears an interruption, detects a voicemail.
MeasurementResponse time, calls handled without a person, conversions, and what to fix next.
Google CloudModels and compute
ModelsGemini 3.8 Live speech to speech, or speech to text, Gemini 3.8 Flash and text to speech in sequence.
Agent runtimeGemini Enterprise Agent Platform: the runtime and its security boundaries.
Your projectPer-customer isolation. Recordings and data stay in your storage.
Integrations
TelephonyTwilio, Telnyx, Plivo, Exotel, Genesys, SIP trunks
CRMSalesforce, HubSpot, Zoho CRM, KaptureCX
SchedulingCal.com, Calendly, Google Calendar
TicketingZendesk, Freshdesk, KaptureCX 360, Re:amaze
Messaging and operationsWhatsApp, SMS, web chat, Slack, Gmail, Zapier
E-commerce and customShopify, Google Sheets, MCP Filesystem, custom REST APIs

Anything not on this list, including your shipping system, courier partners, marketplace seller tools or booking engine, is connected through its API during setup. Tilicho builds and maintains it at no extra cost.

The models

Gemini 3.8 LiveSpeech to speech. One model listens, reasons and speaks in real time. This is the model on the call.
Gemini 3.8 FlashReasoning in the cascaded pipeline, and the analysis after each call.
Chirp 3 and Gemini 3.8 Flash TTSSpeech to text and text to speech for the cascaded pipeline.
Gemini Enterprise Agent PlatformThe agent runtime and its security boundaries, inside your own Google Cloud project.
Questions

Common questions

Which businesses are these agents for?

Online retailers and D2C brands, marketplaces and their sellers, quick commerce, online travel agencies, airlines, hotel groups, tour and holiday operators, and bus, rail and ferry booking platforms.

Which languages does the agent speak?

More than 25 languages. In Southeast Asia: Bahasa Indonesia, Malay, Thai, Vietnamese, Filipino and Taglish, English, Mandarin and Tamil. The agent switches language during the call when the customer does, including mixed speech such as Malay with English.

Can one agent serve several countries?

Yes. The same agent runs on your numbers in each country, with the language, currency, return policy and calling hours set per country.

How do the calls reduce failed cash-on-delivery orders?

Cash on delivery is common in Indonesia, Vietnam, the Philippines and Thailand. The agent confirms each COD order and its address before dispatch, so orders the customer did not mean to place are cancelled before they ship. When a delivery fails, it calls the same day and books a new date or a corrected address, so the parcel is attempted again instead of going back.

Does the agent take payments on the call?

No. It never asks for a card number, OTP or PIN. It sends a payment link from your payment gateway on WhatsApp or SMS, for bank transfer, e-wallet or card, and confirms on the call when the payment lands.

What happens when the customer wants a person?

The call is transferred live to your team with the order or booking, the history and the summary, so the customer does not repeat the case. You decide which cases always go to a person, such as fraud concerns, refund disputes and travellers stuck at an airport.

Where are recordings and customer data kept?

In your own Google Cloud project and your storage, in the region you choose. Tilicho runs the agent inside that project.

Do we have to change our store, shipping or booking system?

No. The agent connects to your existing numbers and systems. There are more than 50 integrations, including Shopify, and other systems, such as your shipping system or booking engine, are connected through their APIs during setup.

See it on one of your use cases

A 30-minute session: a live call with the agent, then a walk through one use case of yours, such as COD confirmation, failed-delivery calls or a flight schedule change.

Write to
neelmohan@tilicho.in