Answer customer questions on WhatsApp with an AI Agent trained on your business. Resolve routine requests, check live information, and hand complex cases to your team.
Official WhatsApp APINo-code setupReturns & exchangesIllustrative demo
ChatSyncs Support
online · AI customer support
The shoes I received are too small. Can I exchange them?
11:08Size exchanges are available within 7 days for unworn shoes with box and tags. What’s your order number and preferred size?
11:08#EX-2051 arrived yesterday. Unworn, with box and tags. I need UK 8 instead of UK 7.
11:09Our returns team will confirm stock and pickup here. Please keep the box and tags.
Size exchange · #EX-2051
Request sent to returns team
UK 7 → UK 8 · Awaiting team confirmation.
Answer routine requests from your business knowledge and give complex conversations to the right teammate.
Recognize customer intent, keep track of the conversation, and choose whether to answer, look something up, or escalate.
Use FAQs, policies, documents, product images, spreadsheets, and API knowledge together for relevant answers.
Connect HTTP APIs to check orders, delivery, CRM records, or tickets, then respond using the returned information.
Assign a teammate, distribute conversations across a team role, or transfer to another AI specialist.
Add labels, save customer details, start or stop follow-up sequences, and trigger your existing bot flows.
Mark resolved conversations as solved, leave internal notes, set staff reminders, skip irrelevant replies, and block clear spam.
These example workflows show how knowledge, live tools, and human handover work together. Adapt the conditions and routing to your own support process.
“Can I return an unopened item?”
##trigger_bot_flow## : 15##mark_as_solved##The customer gets a relevant answer and a next step, with resolution tracked in the inbox.
“My order is late. Can you check it?”
##save_custom_field## : 42##call_http_api## : 5 | reply_with_response | status, message##assign_human_role## : 3Live information guides the reply, and a teammate receives the context if further help is needed.
“I need to speak with your billing team.”
##add_label## : 12##add_note## : Billing team requested##follow_up## : 24The customer does not have to start again. Your teammate can read the conversation and its internal notes.
Type ## inside the System Prompt to find an action. Select your label, sequence, field, API, flow, store, or teammate, then describe when the agent should use it.
The editor displays readable resource names and stores their IDs. The IDs below are examples. Choose resources from your own ChatSyncs account; a label ID, an API ID, and a team ID refer to different things.
Turn information from the conversation into an organized contact record.
##add_label## : 12Apply a label you have created to identify a customer’s interest or the type of request.
Example: Label a pricing enquiry “Interested Buyer” or an unresolved case “Needs Review”.
##remove_label## : 12Remove an existing label when it no longer describes the customer or conversation.
Example: Remove “Needs Review” after your team resolves the issue.
##save_custom_field## : 42Capture an answer in a selected field on the contact, ready for your team or the next automation.
Example: Save the order number, preferred language, requested quantity, or service requirement.
Connect an open-ended WhatsApp conversation to your structured automations.
##assign_sequence## : 8Enroll the contact in a configured sequence when the condition in your prompt is met.
Example: Start a sales follow-up after a customer requests a quote.
##remove_sequence## : 8Remove a contact from a sequence that is no longer appropriate.
Example: Stop a sales follow-up when the conversation has moved to support.
##trigger_bot_flow## : 15Trigger an existing visual bot flow or postback for a process with defined steps.
Example: Launch a booking flow after collecting the service, preferred date, and time.
Use a connected API when the agent needs current information from another system.
##call_http_api## : 5 | reply_with_response | status, messageCall the selected API and form a customer reply from the chosen response fields.
Example: Read an order’s status and delivery message from your logistics system.
##call_http_api## : 5 | reply_with_static_replyRun the API action and use the customer reply you have configured for that action.
Example: Send an acknowledgement after submitting a service request.
##call_http_api## : 5 | no_replyCall the connected system without adding a customer-facing message.
Example: Update your CRM in the background while the conversation continues.
Give each request to the person or AI specialist best equipped to handle it.
##assign_human_member## : 335Send the conversation to a selected member of your team.
Example: Route a key account’s request to their account manager.
##assign_human_role## : 3Distribute a conversation across the members of a configured role using round-robin assignment.
Example: Send a technical issue to your Support role instead of one fixed person.
##transfer_agent## : 18Pass the conversation to another configured specialist when the task changes.
Example: Move a qualified lead from the Qualification Agent to the Sales Agent.
Keep your human team informed and make the next step easy to follow.
##add_note## : Customer wants a callbackWrite context in the shared inbox for your team. The note is not sent to the customer.
Example: Record what the customer tried, the order involved, and the help they need.
##follow_up## : 24Create an internal follow-up reminder after the specified number of hours.
Example: Remind a teammate to revisit an unresolved request in 24 hours.
##mark_as_solved##Close the conversation once the configured resolution conditions are satisfied.
Example: Mark a how-to enquiry solved after the customer confirms the answer helped.
Tell the agent when to pause, stay silent, or protect the inbox from clear spam.
##skip_reply##Avoid replying to irrelevant or vague content when your instructions call for silence.
Example: Do not send a new answer to a message that adds no actionable request.
##block_subscriber##Block a contact when the message matches the spam conditions you explicitly define.
Example: Handle repeated scam messages. A frustrated customer should be routed to support.
An agent can also search products, look up orders, update Shopify order tags or WooCommerce order notes, and cancel orders. Select the connected store first. Product images belong in Knowledge Sources → Media, where the agent can find and send the relevant photo.
Labels describe the contact, custom fields hold the details, and sequences or bot flows move the process forward. Combine them in the agent’s instructions.
When to apply itThe customer requests a quote or confirms purchase interest.
What happens nextSave their requirements and assign a sales follow-up sequence.
When to apply itAn order exception or unresolved question needs your team.
What happens nextAdd context in an internal note and assign the Support role.
When to apply itThe customer asks about a larger quantity or wholesale order.
What happens nextSave the quantity in a custom field and transfer to a sales specialist.
When to apply itA required detail is still missing from the conversation.
What happens nextAsk one clear question and wait before running the next action.
These are suggested label names, not preset labels. Create the names that suit your business, select them in the prompt editor, and define the condition for adding or removing each one.
A staff reminder and a message sequence do different jobs. ##follow_up## reminds your team; ##assign_sequence## enrolls a contact in the follow-up sequence you configured.
Use the information you already have. Give each specialist its own knowledge set, or share the same set across your WhatsApp AI team.
Add business information directly as text: support answers, product policies, service details, or sales playbooks. Useful for information that does not need its own document.
Example: delivery areas, opening hours, and your standard returns policy.
Import a help article, pricing page, or another useful webpage. Select the relevant content and exclude unrelated navigation or page sections when preparing the source.
Example: the troubleshooting steps from a product help page.
Upload reference documents or supply a file URL. Supported source formats include PDF, Word, Excel, CSV, OpenDocument, and plain text.
Example: a product guide, service brochure, or detailed support handbook.
Add image uploads or URLs with descriptions such as product name, color, size, and price. The agent can send the relevant saved images during a conversation.
Add photos here rather than putting image URLs into the System Prompt.
Import a tab from your connected Google account as a structured knowledge source. Use clear columns so product, pricing, and FAQ information is easy to maintain.
Example: a catalog with SKU, product name, size, price, and availability columns.
Import a connected HTTP API’s data into a knowledge set. This supplies background information that can be used with your other sources.
For a fresh lookup during the conversation, use the separate ##call_http_api## action.
Update the knowledge set your agents use without rewriting every prompt.
A useful prompt is a working brief: the agent’s role, the knowledge it should use, the information to collect, and the conditions for each action.
Give the agent one clear job. Describe how it should communicate with customers on WhatsApp.
Identify the sources it should rely on. Tell it which question to ask when a lookup needs an order number or another detail.
Explain what must happen before each action runs. Place the selected action alongside that instruction.
Define when to transfer to another specialist or assign a human. Include the context your team needs.
Explain when the task is finished, when to mark it solved, and what the agent should avoid guessing or promising.
Use your knowledge for routine questions. Collect the useful context before routing a case that needs a person.
You are our WhatsApp Customer Support Agent. Be clear, friendly, and concise. Answer using the attached support knowledge. Ask for missing information before attempting a lookup. When the customer supplies their order number, save it: ##save_custom_field## : 42 When an order-status lookup is needed, call the configured API: ##call_http_api## : 5 | reply_with_response | status, message For a billing dispute, an unresolved issue, or a request for a person: ##add_label## : 12 ##add_note## : Customer needs specialist support; review the conversation ##assign_human_role## : 3 ##follow_up## : 24 Only mark the conversation solved when the issue is resolved: ##mark_as_solved## Do not guess a refund outcome or promise an unsupported resolution.
Adapt this example to your policies and connected tools. Replace every sample ID using autocomplete before saving. The examples describe workflows you configure; they are not preconfigured agents.
Connect your WhatsApp Business number. Add support documents, FAQs, and the systems your team uses.
Write plain-language instructions for answers, live lookups, and escalation. Choose the actions and team members the agent can use.
Try real support scenarios, activate the agent, and improve its knowledge and routing as your needs change.
A product question can become an order enquiry or a support request. Route each message to a specialist with its own instructions, knowledge, and actions.
Only the agents you enable take part in routing. A selected specialist stays with its task until it completes the work, transfers the conversation, or assigns a person. Configure the memory window so agents can use earlier messages, selected products, order numbers, and saved customer information.
Your WhatsApp AI team“Which product is right for me?”
Product knowledge · Recommendations
“Where is order #1042?”
Live order lookup · Delivery information
“Can someone help with my order?”
Business policies · Human handover
Have a question about your WhatsApp AI Agent? Our team can help with setup, routing, and integrations.
Chat on WhatsAppStart with customer support, or connect your store for sales and order assistance. Every agent works inside WhatsApp.
CommerceTurn store conversations into sales with live Shopify data.
CommerceYour WooCommerce sales and support desk, inside WhatsApp.
Build AI Agents that answer your customers, support your store, and keep your WhatsApp conversations moving.