It's 11 PM. A potential customer messages your WhatsApp number: "Hi, do you deliver to Noida?" Your team is asleep. By morning, the customer has already ordered from someone who answered within seconds.
This plays out thousands of times every day across Indian businesses. WhatsApp is a 24-hour channel, but most businesses are not 24-hour operations. The gap between when customers expect a reply and when businesses can provide one is where sales go to die.
Auto-reply fills that gap. But not the generic "We'll get back to you soon" kind. The kind that actually answers the question, greets new contacts with your brand's voice, and hands off to a human when the conversation needs one.
Before Auto-Reply: What Most Businesses Do
The WhatsApp Business app offers a basic away message — one static text that fires whenever you are not available. It helps, barely. The customer sees "We are currently away and will respond shortly" and thinks: every other business sends this exact same message. It is the digital equivalent of an answering machine nobody checks.
Some businesses hire a night-shift person to handle WhatsApp. That works until the night-shift person is sick, or the volume spikes during a festival campaign, or the person gives inconsistent answers because there is no standardized response library.
Enterprise-grade chatbot platforms solve the problem, but at enterprise prices. Setup fees of ₹15,000 to ₹35,000, monthly costs of ₹8,000 and up, and integration complexity that requires a developer on payroll. For a salon, a coaching centre or a local restaurant, that is not a business case — it is a budget line item that never gets approved.
After Auto-Reply: What Changes
With LigiComms auto-reply on your WhatsApp number, here's what happens when that 11 PM customer messages "Hi, do you deliver to Noida?"
- The system checks if this is a new conversation (the customer hasn't messaged within the current 24-hour session). If it is, the welcome trigger fires — sending a greeting with your business name, a line about what you do, and quick-reply buttons for common questions.
- The keyword engine scans the message for trigger words. "Delivery" or "Noida" matches a keyword rule that sends your delivery zone information — areas covered, delivery times, minimum order value — as a structured response with an image header and footer.
- If no keyword matched (say the customer asked something unusual), the AI fallback trigger composes a response from your business notes ("We deliver to all NCR locations including Noida, Ghaziabad, and Gurugram. Free delivery on orders above ₹500.") — written in the customer's own language, using conversation history for context.
- If the customer types "agent" or hits the maximum bot-turn count, human handoff kicks in — the auto-reply stops, a handoff message is sent, and the conversation is flagged in the Sociobox inbox for human attention.
The customer got their answer in under two seconds. Your team wakes up to a flagged inbox of only the conversations that needed a human — not the 40 messages asking "what time do you open?"
Auto-reply isn't about replacing your team. It's about handling the 80 percent of messages that have a known answer so your team can focus on the 20 percent that need judgment.
The Trigger System: Welcome, Keywords and AI
Auto-reply rules are evaluated in priority order, and each rule can be scoped to specific hours:
Welcome Trigger
Fires when a customer opens a new conversation — bounded by WhatsApp's own 24-hour session window. A contact who messaged three months ago and writes again today gets the welcome. A contact who already messaged an hour ago does not. The boundary is WhatsApp's session window, not a fixed timer — so it resets naturally.
Keyword Triggers
Each rule carries a keyword and a response. "Price" sends the price list. "Hours" sends opening hours. "Return" sends the return policy. Keywords are matched case-insensitively by whole-message normalised comparison — "stop by tomorrow" does not trigger the "stop" opt-out keyword. Each trigger can run only during business hours, only after hours, or any time.
AI Fallback
When no keyword matches, the AI composes a response based on business notes you provide, the customer's message, and the last few conversation turns. The AI generates text in the customer's own language — if they wrote in Hindi, the reply comes in Hindi. A hard timeout prevents a slow model response from stalling the webhook. If the AI fails, the conversation simply waits for a human.
Guardrails That Prevent Runaway Bots
The biggest fear with auto-reply is a bot that won't shut up — sending message after message in a loop, annoying the customer, or racking up API charges. LigiComms has multiple layers of protection:
- Per-contact cooldown. After a reply, the same contact cannot trigger another auto-reply within a configurable cooldown window. This prevents rapid-fire responses to a customer who sends five messages in ten seconds.
- Consecutive bot-turn cap. After a configurable number of bot replies without a human intervening, auto-reply stops and escalation fires. The bot knows when it is out of its depth.
- Daily cap per contact. A maximum number of auto-replies per contact per day, derived from stored messages — no counter to drift or reset.
- Human takeover detection. If a human team member replies to a conversation, auto-reply pauses for that conversation. The bot recognizes its colleague. Each outbound message records whether a human or the bot sent it, so the system never mistakes its own reply for a human's.
- Atomic claim. Before any reply is sent — before AI is called, before rate limits are checked — a single atomic database operation claims a reply slot. If two webhook deliveries race for the same contact, only one wins. This prevents duplicate replies, duplicate AI calls, and duplicate charges.
Opt-Out That Works Unconditionally
If a customer sends "stop," they are opted out. Period. This does not depend on the tenant's subscription plan, the feature being enabled, or the auto-reply configuration. Opt-out is honoured unconditionally because consent is not a premium feature.
The opt-out writes to the same contact consent system that campaigns use, so a customer who opts out of auto-reply is also opted out of broadcast templates. One "stop" means stop — everywhere.
Opt-in works the same way in reverse: "start" re-enables messaging for that contact.
What This Costs
Auto-reply is included on every paid plan (Basic, Pro, Premium). There is no separate chatbot add-on, no per-agent seat cost, no per-message fee beyond Meta's standard conversation rate. Configure your rules, enable the feature, and it runs on your own WhatsApp Business number — the same one you connected via Embedded Signup.
AI-generated replies use the same AI rate limits as caption generation (30 per tenant per hour). For most businesses, that is more than enough for after-hours traffic. High-volume businesses are better served by keyword rules for common queries and AI for the long tail.
When Auto-Reply Pays for Itself
A salon that answers "What are your hours?" automatically saves one front-desk employee fifteen minutes per day. A clinic that confirms appointments saves a full-time assistant's worth of follow-up calls per week. A D2C brand that responds to "Where is my order?" with tracking information at midnight saves the sale that would have been cancelled by morning.
Auto-reply is not a chatbot gimmick. It is operational infrastructure. And when it runs on your own number, inside your own dashboard, with your own rules — it is the employee that never sleeps, never calls in sick, and never gives an answer you didn't approve.