FLOW · visual automation

Drag, drop,
automate your replies.

Lane.Chat Flow turns the "if a customer asks X, do Y" logic in your head into a visual DAG — triggers, branches, AI replies, API calls, handoffs. Ship it in an afternoon, not a sprint.

Build your first flow Open the dashboard
8 node types drag-and-drop visual canvas 0 lines of code required
NODE TYPES

Eight node types. Endless combinations.

Each node does one job well. String them together and you've got anything from a simple FAQ bot to a full refund + escalation workflow.

ENTRY

Trigger

Kick the flow off from a message, a keyword, a button, a webhook, a schedule — or a customer hitting a specific page.

keyword intent webhook schedule
LOGIC

Branch

Split on anything — the customer's message, a variable, an API response, the time of day, their tier.

if / else switch regex match
ACTION · AI

AI reply

Let AI answer with your knowledge base, persona and brand tone. Falls back to handover if confidence drops.

GPT · Claude · Gemini RAG from your KB confidence threshold
ACTION · HUMAN

Handover

Pass the thread to a live agent — with context, tags and the conversation so far already attached.

team routing round-robin priority queue
ACTION · MESSAGE

Send a message

Push a message, template, rich card or button set back to the customer on any channel they're on.

text · image · video buttons & quick replies templates
ACTION · DATA

Tag & update

Add tags, set variables, update the contact's profile. Feeds filters, reporting and the next time they come back.

contact tags custom fields segment
ACTION · API

Call an API

Hit any endpoint — your order system, Stripe, Shopify, internal tools. Use the response in the next node.

REST Webhook headers & auth JSON path
FLOW CONTROL

Wait

Pause for 5 minutes, a day, or until the customer replies — then pick up where you left off.

duration until reply until event
CANVAS

A real flow, not a toy you can poke at.

Real flow from a Shopify store running on Lane.Chat. Drop new nodes, wire them up, hit publish — no redeploy, no YAML files to push.

flow · refund-auto.yaml
6 nodes 7 edges Last run · 2 min ago
Trigger
Customer says "refund"
When a customer message contains "refund" on Telegram, Web, or WhatsApp.
Branch
Auto-refund eligible?
Order total < $50 AND placed within 7 days.
API
Refund through Stripe
POST /api/refund/auto
Handover
Hand to a human
Route to the on-shift agent with full order context.
Message
Send "refund processed" reply
Confirmation + receipt back to the customer.
Tag
Tag for follow-up
Tag as "refund-manual" for the billing team.
yes
no
12.4k
triggers this month
81%
handled without a human
2.4s
average resolution
0.3%
hit the human queue
IN THE WILD

Flows people actually ship.

Three flows running on live Lane.Chat accounts today. Copy the idea, swap the details, you're off.

Case 01 · Refund flow

Refunds in two seconds, not tomorrow.

Small refund, recent order, customer on Telegram: done in 2 seconds, no ticket opened. Anything borderline gets a human — with the order already pulled up.

81% auto-resolved
2.4s average response
-62% ticket volume
Keyword trigger"I want a refund"
Check eligibilityAmount & order age
Call refund APIPOST /refund/auto
Reply to customerConfirmation + receipt
Case 02 · Booking flow

Booking that doesn't need your day off.

Customer asks for a slot at 2am? Flow checks the calendar, offers three times, confirms, and schedules a reminder. Your team wakes up to a full calendar.

24/7 bookings taken
+38% after-hours capture
0 double-books
Intent trigger"Can I book a time?"
Fetch open slotsGET /calendar/slots
AI confirms slotOffer three options
Wait until booking day24h before the slot
Tag as "scheduled"Reminder sent
Case 03 · Post-purchase check-in

Follow-ups that don't feel like spam.

Three days after someone buys, ping them. Ask how it's going. If the vibe is off, loop in a human. If it's great, ask for a review. Simple, measurable, kind.

+41% review rate
+312 new reviews / mo
18 issues caught early
Scheduled trigger3 days after purchase
Fetch orderPull order details
AI checks sentimentHappy · meh · unhappy
Wait for answerReply or timeout
Tag & branchRoute to review or human
The best customer-service automation
is the kind your team can change on Tuesday without filing a ticket.
Why we built Flow the way we did
FLOW + AI

AI is the muscle. Flow is the brain.

Most "AI chatbots" are a single prompt praying for a miracle. Flow is the scaffolding around AI that makes it actually work in production.

Flow ORCHESTRATES
Profile & history
Knowledge base
Agent handoff
Conversations
1

Flow feeds AI the right context

The trigger node catches the customer's intent, pulls their profile, their last orders, their tags — AI gets all of it before it thinks.

2

Knowledge base becomes a node

Ask the RAG node to pull the refund policy, the latest promo, the product spec. Flow injects it into the prompt, AI answers from facts.

3

Humans stay in the loop, on purpose

Low confidence, angry customer, VIP tag — Flow reads AI's output and decides whether to send, edit, or escalate to a person.

4

Every run is replayable

Every run logs what fired, what branched, what the API returned. Find a broken flow in seconds, not scrolling through logs.

RELATED

Flow pairs well with

Flow plays nice with the rest of Lane.Chat. Here's where it connects.

Channels

Every channel Flow can trigger on — Web, Telegram, WhatsApp, Slack, Line and more.

See channels →

Roleplay AI

Give the AI reply node a personality — tone, style, boundaries, even a voice.

See roleplay →

Knowledge base

Feed the AI reply node with your docs, help articles, FAQs and product data.

See knowledge →

Workbench

Where handovers land. Agents pick up flows mid-run with full context.

See workbench →
✺ Ready when you are

Stop writing the same reply
for the hundredth time.

Turn the questions you answer every day into flows that run on their own. Humans stay in the loop for the messy stuff.

Start free — build your first flow See pricing

INCLUDED

  • Unlimited nodes on all paid plans
  • Visual builder — no YAML, no deploys
  • Built-in AI, knowledge base and API nodes
  • One-click handoff to a live agent