If you run a clinic, travel agency, coaching centre, salon, or any local service business, your support team probably answers the same questions dozens of times a day: What are your hours? How much does it cost? Where are you located? Is my booking confirmed? Can I talk to someone?

Typing those replies manually does not scale. Customers expect instant answers on WhatsApp — and they leave when they have to wait.

The fix is not hiring more people to copy-paste. It is automating the repetitive parts while keeping humans available for the conversations that actually need empathy and judgment.

In Pyngdesk, support automation works in two layers:

• Quick Replies — keyword-triggered FAQ answers (price, hours, address) that fire instantly without starting a conversation flow.
• Flows — visual conversation paths (menus, text capture, media, lookups, handoff) powered by one live flow at a time. When a customer messages and no Quick Reply matches, Pyngdesk auto-starts your active flow — whether they say hi or describe a full enquiry in free text.

This guide covers five flows every service support team should automate — with concrete examples for clinics, travel, and coaching businesses, and exactly how to build them in Pyngdesk.




Why Automate Support on WhatsApp (Not Email or Phone)

WhatsApp is where your customers already are. For service businesses in India and beyond, it is often the first channel people choose after finding you on Google or Instagram.

Compared to email, WhatsApp messages get read quickly and feel personal. Compared to phone IVR (“Press 1 for billing…”), WhatsApp lets customers reply in their own words and get structured menus when they prefer tapping buttons.

For inbound support, the WhatsApp Business API gives you:

• Instant self-service for FAQs — hours, pricing, location, brochures
• Structured intake before an agent joins — service type, preferred date, notes
• A clear handoff when automation should stop — with full context attached to the chat
• Proactive follow-ups via approved templates when you need to reach out after the 24-hour window (see our guide on running your first WhatsApp campaign safely)

The goal is not to replace your team. It is to stop them from answering “We’re open 9–7” for the hundredth time this week.

How Pyngdesk Automation Works (Quick Primer)

Before diving into the five flows, here is how inbound messages are handled:

Layer	What it does
Quick Replies	Exact keyword match on the whole message (price, hours, brochure) → instant canned answer; no flow session
Live flow	One active flow per org. Any other inbound text (e.g. hi, “I want to book for June”) auto-starts it at the first node
Menus	Button nodes (up to 3 options) and list nodes (up to 10) inside a flow
Data capture	Text nodes with save_as fields (name, date, notes)
Rich answers	Image, document, audio, and location nodes
Lookup	dataset_lookup / dataset_list against uploaded CSV datasets
Multi-choice	multi_select for “pick all services you need”
Wait & nudge	Delay node — send a follow-up after X minutes unless they reply
Escape hatch	End node marked as handoff → routes chat to your agent queue
Smarter routing	AI-enabled flows (Gemini) for messy free text inside an active session
Agent inbox	Chats flagged needs_agent with attention alerts

Important: Keywords belong in Quick Replies, not on flows. Greetings like hi and hello are reserved for starting the live flow — do not add them as Quick Reply keywords.

Flow 1: Welcome & Issue Routing

A customer messages “Hi” at 10 AM. Without automation, someone on your team has to notice the chat, open it, and type “How can I help you?” — often minutes later. With a routing flow set as your live flow, the customer immediately gets a clear menu of what you can help with.

How it starts

What it solves

Activate this flow as your org’s live flow (is_active). Then any inbound message that does not match a Quick Reply keyword starts the flow automatically — including:

• “Hi” / “Hello”
• “I need help”
• “I want to book a Kerala trip in December”

You do not configure trigger words on the flow itself. Keywords are only for Quick Replies (Flow 2).

Recommended node path

1. Set this flow as Active (live) in Flows → Overview
2. Start node:list or button — “What can we help you with today?”
   • Book appointment / enquiry
   • Pricing & packages (or branch to a document node; FAQ keywords like price are handled by Quick Replies when sent alone)
   • Office location
   • Talk to support
3. Branch each option to the right sub-path, rich-media node, or handoff end node

What the customer sees

A WhatsApp list or button message with tappable options — no typing required unless they choose a free-text path.

What the team gains

Fewer “Hi… waits …Hi again?” exchanges. Agents only join when the customer picks “Talk to support” or reaches a handoff node.

Service-business examples

Business	Menu options
Clinic	New patient · Follow-up · Billing · Speak to reception
Travel	Domestic trips · International · Custom itinerary · Agent callback
Coaching	Demo class · Fees & batches · Syllabus · Speak to counsellor

Pro tip

Use a list node when you have more than three options (up to 10). Reserve button nodes for your top three highest-volume intents. Start the Travel Enquiry template in Pyngdesk as a reference — activate it as your live flow; it already demonstrates welcome → destination → brochure → location branching.

---

Flow 2: Hours, Pricing & Location (FAQ Layer)

What it solves

A large share of inbound chats ask the same four facts: opening hours, price list, office address, and “send me your brochure.” These do not need a human — but they do need accurate, consistent answers.

Example customer messages (Quick Reply keywords)

Quick Replies use exact match on the whole message (case-insensitive). The customer must send essentially one keyword or short phrase:

• hours or timings → office hours reply
• price or how much → pricing reply
• address or location → address reply
• brochure or pdf → document reply

Do not use greetings (hi, hello) as Quick Reply keywords — those should start your live flow (Flow 1).

Recommended build — two layers

Layer 1: Quick Replies (fastest win)

In Quick Replies, create keyword-triggered responses:

Quick Reply	Keywords	Response
Office / support hours	hours, timings, open, when are you open	Mon–Sat 9:30 AM – 7:00 PM IST…
Pricing overview	price, cost, fees, how much	Your standard rate card (text or PDF)
Office address	address, location, where are you	Short address + “Reply location for map pin”
Company brochure	brochure, pdf, catalogue	Document attachment with a one-line caption

Quick Replies run before the live flow: if price matches, the customer gets the rate card and no flow session starts. They work on idle chats and, for hours/address only, can be allowed during agent handoff via allow_during_handoff.

Layer 2: Flow nodes for richer answers

• location node — map pin with name and full address for “Where are you?”
• document node — fee structure PDF or course catalogue
• image node — clinic reception, campus, or tour photos with caption

Wire the “Office location” branch from Flow 1 to a location node, and “Pricing” to a document or text summary node.

What the customer sees

Instant answers — text, PDF, image, or a tappable map pin — without waiting in a queue.

What the team gains

Your agents stop being a human FAQ page. Set quick-reply priority so pricing and hours beat generic fallbacks.

Pro tip

Enable allow_during_handoff only for harmless facts (hours, address) so customers can still get basics while waiting for an agent. Do not auto-reply with marketing promos during handoff — it feels tone-deaf.

---

Flow 3: Structured Enquiry / Appointment Intake

What it solves

When a customer wants to book a consultation, trip, or demo class, agents waste time asking “Which service? What date? What’s your name?” in a back-and-forth that could have been collected upfront.

How customers reach this path

• They message freely (“I want to book an appointment”) → live flow starts → they pick Book appointment from the welcome menu, or
• They are already in an active flow session and selected Book appointment from Flow 1’s list

Recommended node path

1. List or multi_select — “Which service are you interested in?”
   • Clinic: cleaning, root canal, whitening, consultation
   • Travel: Kerala, Goa, Maldives, custom
   • Coaching: Grade 10 batch, Grade 12 batch, crash course
2. Text node — save_as: preferred_date — “What date works for you?”
3. Text node — save_as: customer_name — “May we have your name?”
4. Text node — save_as: notes — “Anything else we should know?” (optional)
5. Delay node (optional) — “Thanks! We’ll confirm within 2 hours.” with cancel_on_reply so a follow-up message sends only if they go quiet
6. End node — summary: “We received your enquiry for {{service}} on {{preferred_date}}. We’ll call you shortly.”

If the enquiry needs a human, branch to Flow 5’s handoff end instead of a plain end.

What the customer sees

A guided conversation — one question at a time — that feels like a receptionist, not a form.

What the team gains

When an agent opens the chat, session answers already show service, date, and name. No re-asking.

Pro tip

Use multi_select when customers commonly need more than one service (e.g. “teeth cleaning + whitening”). For single-choice menus, a list node is simpler. Model your flow on Pyngdesk’s Travel Enquiry template: destination → dates → budget → handoff.

---

Flow 4: Booking or Reference Status Lookup

What it solves

“Is my appointment confirmed?” and “What’s my booking status?” are high-volume, low-complexity questions — if you have structured data. Agents should not look up the same spreadsheet fifty times a day.

How customers reach this path

Best wired as a branch inside your live flow (e.g. welcome menu → Check booking status), or reached after the live flow auto-starts and the customer navigates there. Optionally add a dedicated Quick Reply keyword like status if you want a one-word entry that skips the welcome menu.

Recommended node path

1. Upload a Dataset (CSV) with columns such as:
   • booking_id, customer_phone, status, appointment_date, service_name
2. Text node — save_as: lookup_ref — “Please share your booking reference or registered mobile number.”
3. dataset_lookup node — match on booking_id or customer_phone per your config.match setting
4. Branches:
   • Found → text node: “Your {{service_name}} on {{appointment_date}} is {{status}}.”
   • Not found → clarify once, then offer handoff: “We couldn’t find that reference. Reply agent to speak with our team.”

What the customer sees

Self-serve status in seconds — no hold music, no “let me check and get back to you.”

What the team gains

Deflection of routine status checks. Agents handle exceptions only.

Service-business examples

Business	Lookup key	Sample status values
Travel	Booking reference	Confirmed · Pending payment · Cancelled
Clinic	Registered mobile	Scheduled · Completed · Rescheduled
Coaching	Enrollment ID	Active · Waitlisted · Payment due

Pro tip

Keep your dataset fresh — stale data erodes trust faster than no automation. Export from your practice-management or booking tool weekly, or after every batch of confirmations. Test with three known records in the flow simulator before going live.

---

Flow 5: Smart Handoff to a Human (With Context)

What it solves

Automation should know when to stop. Customers who are upset, confused, or explicitly ask for a person should reach a human quickly — with everything the bot already collected attached to the chat.

When handoff fires

• Customer picks Talk to support (or similar) from a menu inside the live flow
• Customer fails clarification 3 times (Pyngdesk’s default clarify_max_attempts) while in an active session
• AI detects high-stakes intent: refund, complaint, cancellation (when AI is enabled on the flow)

Note: while a chat is handed off to an agent, new messages do not start the live flow again. Only Quick Replies with allow_during_handoff (e.g. hours) may still auto-respond.

Recommended node path

1. Button or list option — “Talk to agent” → routes to handoff end node
2. End node — mark as handoff (is_handoff: true in node config, or key like end_handoff)
3. Engine calls handoff_session — chat status becomes needs_agent
4. Agent sees the chat in Chats with attention alert; session answers from Flow 3 are visible

Optional enhancements:

• write_dataset_log on the flow — append every completed intake to a dataset for reporting (“47 enquiries this week, 12 handed off”)
• AI instructions on the flow: “Hand off immediately for refunds, complaints, and medical emergencies.”

What the customer sees

“Connecting you with our team now. An agent will reply shortly.” — then a real person who already knows their name, service, and date.

What the team gains

No cold handoffs. Faster resolution. Clear queue (needs_agent) so nothing slips through during busy hours.

Pro tip

Pair handoff with Flow 2’s hours quick reply: outside business hours, the handoff end message should set expectations — “Our team is offline until 9:30 AM. We’ve noted your request and will reply first thing.”

---

Which Flow Should You Build First?

Not every team needs all five on day one. Use this rollout order:

Order	Flow	Effort	Impact
1	FAQ Quick Replies (Flow 2)	Low	High — immediate deflection
2	Welcome & routing (Flow 1)	Medium	High — structures every new chat
3	Enquiry intake (Flow 3)	Medium	High — better agent productivity
4	Handoff with context (Flow 5)	Low	Critical — safety net for all flows
5	Dataset lookup (Flow 4)	Medium–High	Medium — needs clean booking data

Ship Quick Replies this afternoon. Add routing by end of week. Layer intake and handoff before you upload your first dataset.

---

Common Mistakes to Avoid

1. Putting hi in Quick Replies — Greetings should start your live flow, not bypass it with a canned reply.
2. No handoff path — Every flow needs a way to reach a human. Customers who feel trapped report spam or block you.
3. Too many questions upfront — Collect three fields, not ten. Answer something useful before asking for more.
4. No after-hours messaging — Pair Flow 1 with a hours Quick Reply or a time-aware welcome message so midnight enquiries get a response.
5. Stale datasets — A lookup that says “Confirmed” when the appointment was cancelled is worse than no lookup.
6. Ignoring the needs_agent queue — Handoff without agent response trains customers to stop using WhatsApp. Monitor Chats and set internal SLAs.
7. Automating complaints — Route refunds, disputes, and sensitive issues to humans via AI instructions or a dedicated “Billing issue” → handoff branch.

---

Getting Started in Pyngdesk

1. Add Quick Replies first for your top FAQ keywords (price, hours, address, brochure) from last month’s chat history.
2. Open Flows and start from a template (e.g. Travel Enquiry) or use the AI flow builder to describe your business in plain English.
3. Test with the built-in simulator, then activate exactly one flow as your live flow.
4. Open Chats and watch handoff volume for the first week. Tune clarify_max_attempts and menu labels based on what customers actually tap.

If you already use the WhatsApp Business app on the same number, Pyngdesk supports coexistence onboarding — your team can keep the mobile app while API flows handle automation. See the project README for webhook subscription details.

FAQ: