Cart recovery bot
Build time: 12 min
Recovers 18-24% of abandoned carts
Visual builder · 40+ nodes · Zero code
Build a WhatsApp bot the way you'd sketch it on a whiteboard.
Drag a trigger. Drop a few message nodes. Fork on a condition. Push to live in one click. No code, no scripts, no sandboxes — your real WhatsApp number, instantly.
Official Meta Tech Partner · Trusted by 500+ businesses worldwide
flow: cart-recovery.v3● live
Trigger
cart_abandoned
Wait
30 min
Message
Hey {{name}} 👋
Condition
cart > ₹2,000
AI Reply
objection handler
CRM Sync
HubSpot
THE NODE PALETTE
40+ building blocks. Snap them together however you want.
Message
Condition
AI Reply
Webhook
Delay
CRM Sync
Trigger
Button
4 STEPS TO LIVE
From blank canvas to a real bot answering your real customers.
PROVEN PATTERNS
Don't start from a blank canvas. Fork a pattern that already works.
COD-to-prepaid converter
Build time: 18 min
Cuts RTO losses by up to 40%
Lead qualifier + site-visit booker
Build time: 25 min
Books 3x more visits than forms
Course counsellor + payment link
Build time: 30 min
Closes leads in <6 hours
Appointment booking + reminders
Build time: 15 min
Reduces no-shows by 35%
Onboarding drip + activation nudge
Build time: 22 min
Lifts D7 retention by 12%
Why teams pick LandinChat over generic builders
| Capability | LandinChat | Most builders |
|---|---|---|
| Native AI fallback node | ✓ Built-in (GPT, Gemini, Claude) | Bring-your-own webhook |
| Versioning + rollback | ✓ Per-flow git-style history | Overwrite live |
| Live preview on real number | ✓ Yes, no sandbox | Test environment only |
| CRM sync nodes | ✓ HubSpot, Zoho, Salesforce, Sheets | Zapier-only |
| Human handoff routing | ✓ By team, hours, language | Single inbox |
| Conversation analytics | ✓ Per-node drop-off heatmap | Aggregate counts |
Builder FAQ
Do I need to code to build a WhatsApp chatbot?
No. The visual builder is fully drag-and-drop. Developers can call webhooks or run JS inside any node for advanced logic.
How long does it take to deploy?
Most teams ship their first bot in 20–30 minutes. Multi-branch flows with CRM sync take 2–4 hours.
Can the chatbot hand off to a human agent?
Yes — drop a Handoff node and route by team, working hours, language, or any variable. Chats land in the shared inbox instantly.
Does it work with the official WhatsApp Business API?
Yes. LandinChat is an Official Meta Tech Partner. The builder runs on the Cloud API directly — no third-party BSP markup.
Can I A/B test flows?
Yes. Fork any node into two branches with a percentage split. Built-in analytics shows conversion per branch.
Your first bot. Live in 30 minutes.
Drop the trigger. Wire the flow. Hit publish. That's it.
Open the builderRule-based, AI, hybrid — pick the right shape
| Bot type | Best for | Risk |
|---|---|---|
| Rule / menu-based | FAQ deflection, order lookup, appointment booking | Feels stiff for open-ended queries |
| AI (LLM) | Open-ended support, product Q&A, discovery | Hallucination, cost per message |
| Hybrid (default) | Menu for known intents, AI fallback for the tail | Requires careful intent routing |
Most production bots on WhatsApp are hybrid: 6-10 menu intents cover 80% of traffic; an LLM handles the tail with strict grounding to your knowledge base.
Anatomy of a bot that ships
- Entry: a keyword, a CTWA ad, an inbound "hi", or a workflow trigger.
- Router: intent classifier (rule + LLM), picks a flow.
- Flow: ordered steps — question, capture, validate, API call.
- State: persistent across the 24h session and beyond.
- Handoff: to a human agent, with full transcript.
- Fallback: "I didn't get that" with a menu, not silence.
5 recipes with clear ROI
- Order status: capture order ID → API lookup → status template. Deflects 40-60% of support volume.
- Appointment booking: service → date → time slot → confirmation. Doubles booking conversion vs form.
- Lead qualifier: 3 questions → BANT score → route to rep or nurture.
- Product finder: quiz → LLM ranks catalog → 3 product cards.
- Renewal upsell: renewal date lookup → tier-up template → payment link.
Handoff to human — the hard part
A bot loses trust the moment a user asks for a human and gets another menu. Every bot needs an explicit escape: keywords ("agent", "human", "support"), a menu option on every fallback, and business-hours-aware routing (queue when reps are offline). On handoff, pass full transcript + captured fields so the agent doesn't re-ask.
AI guardrails on WhatsApp
- Ground the LLM to your KB — reject answers with no source.
- Cap output length (WhatsApp reads short).
- Block price / SLA / legal claims from the LLM path — route to human.
- Log every AI turn with prompt + response for audit.
- Rate-limit per contact to control cost.
Launch checklist
- [ ] Every flow has a handoff path.
- [ ] Every AI turn is grounded + length-capped.
- [ ] Every intent has a fallback with menu.
- [ ] Business hours + timezone respected.
- [ ] Full transcript visible to human on handoff.
Related: Workflow automation · WhatsApp CRM · WhatsApp API.