From "I Wish I Could..." to "Hey Claude, Do This"

Most WhatsApp automation guides show you what's possible. This one shows you what to say.

MCP (Model Context Protocol) connects your AI assistant directly to MoltFlow's WhatsApp tools. Instead of clicking through dashboards, writing API calls, or building n8n workflows, you type a sentence. Claude figures out which tools to call, in what order, with what parameters -- and executes the whole thing.

The five workflows below are not hypothetical. Each one is a prompt you can paste into Claude Desktop or Claude Code right now. They use MoltFlow's 26 MCP tools behind the scenes, but you never touch a single one directly. You just describe what you want.

Every workflow here takes less than 10 minutes to set up and run. Most take less than 2.

Prerequisites

Before running any of these workflows, you need three things:

1. A MoltFlow account with an active WhatsApp session. Sign up at molt.waiflow.app and connect your WhatsApp by scanning the QR code. Your session status should show "working."

2. MCP configured in your AI client. For Claude Desktop, add MoltFlow to your config file:

json

{
  "mcpServers": {
    "moltflow": {
      "url": "https://apiv2.waiflow.app/mcp"
    }
  }
}

For Claude Code, run one command:

bash

claude mcp add moltflow --transport http --url https://apiv2.waiflow.app/mcp

No API key needed -- OAuth handles authentication automatically on first use.

3. Some data to work with. These workflows are most useful when you have monitored groups capturing leads, some chat history, and at least one custom group. But even with a fresh account, workflows 1 and 3 will work as soon as you add a monitored group.

That's it. No npm packages, no local servers, no code to write.

Workflow 1: Morning Lead Digest

The prompt:

"Check all my monitored groups for new leads from the last 24 hours. Summarize them by group, showing each lead's name, phone, score, and what they said. Highlight any leads scored 7 or above."

What happens behind the scenes:

Claude makes two tool calls. First, moltflow_list_monitored_groups retrieves every WhatsApp group you are tracking for lead signals. Then, moltflow_list_leads pulls all captured leads, filtered to the last 24 hours. Claude cross-references the data, groups leads by their source, and formats a briefing.

What you get:

A structured daily digest that looks something like this:

text

--- Lead Digest (March 15, 2026) ---

Property Investors Group (14 new leads)
  * Ahmed K. (+971501234567) - Score: 9/10
    "Looking for 3BR apartments in Marina, budget 2-3M"
  * Sarah M. (+971559876543) - Score: 8/10
    "Anyone know a good agent for JBR rentals?"
  ...

Dubai Business Network (6 new leads)
  * Raj P. (+919876543210) - Score: 6/10
    "Need office space for 20 people, preferably free zone"
  ...

HIGH PRIORITY: 4 leads scored 7+, recommend immediate follow-up.

Why this matters: Without this workflow, you are logging into the dashboard every morning, clicking through each monitored group, and manually scanning for high-value leads. This prompt compresses that into a 15-second read. Run it every morning, or ask Claude to compare today's digest against yesterday's to spot trends.

Variations to try:

"Show me only leads scored 8 or above from the last 48 hours"
"Compare this week's lead volume against last week by group"
"Which monitored group generated the most leads this month?"

Workflow 2: Automated Follow-Up Sequence

The prompt:

"Find all leads with status 'contacted' who were last reached more than 3 days ago. Check their engagement -- if they haven't replied, send each one a follow-up message on my 'sales' session saying: Hi! Just checking in on my earlier message. Would love to help if you're still looking. No pressure at all."

What happens behind the scenes:

This prompt triggers a three-step chain. Claude calls moltflow_list_leads with a status filter to find all leads marked as "contacted." For each lead, it calls moltflow_get_contact_engagement to check whether the contact has replied since the last outreach. For any contact that hasn't responded, Claude calls moltflow_send_message to deliver the follow-up.

What you get:

A hands-free follow-up sweep that catches every lead who slipped through the cracks. Claude reports back with a summary:

text

Follow-up results:
- 12 leads were in 'contacted' status for 3+ days
- 8 had no reply (follow-up sent)
- 4 had already replied (skipped)
- 0 failures

Messages sent to: Ahmed K., Sarah M., Raj P., ...

Why this matters: Follow-up is where most sales pipelines leak. Studies consistently show that 80% of deals require at least five touchpoints, but most people give up after one. This workflow ensures no lead goes cold simply because you forgot. The AI checks engagement data before sending, so contacts who already replied are not bothered with a redundant message.

Variations to try:

"Follow up with leads scored 8+ first, then 5-7 tomorrow"
"Use a different message for leads from the 'Real Estate' group vs. 'Business Network'"
"After sending follow-ups, update each lead's status to 'follow-up-sent'"

Workflow 3: Group Sentiment Monitor

The prompt:

"Pull the last 50 messages from our 'Customer Feedback' monitored group. Analyze the tone and content. Flag any messages that contain complaints, frustration, or requests for help. Summarize the overall sentiment."

What happens behind the scenes:

Claude calls moltflow_get_group_messages to retrieve recent messages from the specified group. The messages come back with content, sender info, and timestamps. Claude then applies its own language understanding -- no separate sentiment API needed -- to classify each message by tone and intent.

What you get:

A sentiment report with specific flagged messages:

text

Sentiment Analysis: Customer Feedback Group
Period: Last 50 messages (March 13-15)

Overall sentiment: Mixed (60% positive, 25% neutral, 15% negative)

FLAGGED - Needs attention:
  1. Omar R. (Mar 14, 2:15 PM): "Been waiting 5 days for
     a response on my refund. This is unacceptable."
     -> Complaint: delayed refund response
  2. Lisa T. (Mar 15, 9:30 AM): "The app crashed twice
     today when I tried to export my data."
     -> Bug report: export feature crash

POSITIVE HIGHLIGHTS:
  - 6 messages praising the new dashboard design
  - 3 messages recommending the product to others

Why this matters: Customer sentiment shifts before churn metrics do. By the time your analytics show a retention drop, frustrated customers have already left. This workflow gives you a real-time pulse on how customers feel. You catch complaints within hours, not weeks. Run it daily, or set up a scheduled prompt that emails you the results every morning.

Variations to try:

"Compare sentiment this week vs. last week -- is it improving or declining?"
"Find messages where someone tagged or mentioned our brand name"
"Pull messages from all monitored groups and rank them by complaint density"

Workflow 4: One-Command Campaign Launch

The prompt:

"List my custom groups, find the one called 'VIP Clients', and create a bulk send to them with this message: 'Exclusive for our VIP members -- 25% off all plans through March 31. Reply INFO for details.' Schedule it for tomorrow at 10:00 AM UTC."

What happens behind the scenes:

Claude chains three tools together. It calls moltflow_list_custom_groups to find the group by name and retrieve its ID. Then it calls moltflow_create_bulk_send with the group ID, message text, and scheduling parameters. Finally, it calls moltflow_create_schedule to set the delivery time. All anti-spam safeguards -- typing simulation, randomized delays between messages, rate limiting -- are applied server-side automatically.

What you get:

A fully configured campaign, ready to go:

text

Campaign created successfully:
- Campaign ID: bs_abc123
- Target group: VIP Clients (47 contacts)
- Scheduled: March 16, 2026 at 10:00 AM UTC
- Message: "Exclusive for our VIP members..."
- Status: SCHEDULED

Estimated delivery window: 10:00 AM - 10:35 AM UTC
(delays applied automatically to avoid spam detection)

Why this matters: Setting up a campaign through any messaging UI takes 5-10 minutes of clicking: select the group, compose the message, set the schedule, review, confirm. Through MCP, the entire flow happens in one sentence. This is especially powerful when you are running multiple campaigns per week. Instead of a 10-minute setup ritual each time, you describe what you want and move on.

Variations to try:

"Create two campaigns: one for VIP Clients at 10 AM and one for Warm Leads at 2 PM"
"Check how my last campaign to this group performed before sending a new one"
"After the campaign runs, check the delivery stats and send me a summary"

Workflow 5: Customer Win-Back

The prompt:

"Look at my recent chats and find contacts who haven't messaged us in the last 30 days. Create a custom group called 'Win-Back March' with those contacts. Then create a bulk send to that group with this message: 'Hey! We noticed it's been a while. We've added some great new features recently. Reply HI and I'll fill you in.'"

What happens behind the scenes:

This is the most complex workflow -- four tool calls chained together. Claude calls moltflow_list_chats to retrieve all recent conversations with their last-activity timestamps. It filters for contacts whose last message is older than 30 days. Then it calls moltflow_create_custom_group to create the "Win-Back March" group with those contacts. Finally, it calls moltflow_create_bulk_send to set up the re-engagement campaign targeting the new group.

What you get:

An automated win-back pipeline from identification to outreach:

text

Win-back analysis complete:
- Total contacts reviewed: 234
- Inactive 30+ days: 38 contacts
- Custom group "Win-Back March" created (38 members)
- Bulk send campaign created: bs_xyz789
- Status: PENDING (ready to send)

Inactive contacts include:
  - David L. (last active: Feb 10)
  - Maria S. (last active: Feb 3)
  - ...

Why this matters: Customer acquisition costs 5-7x more than retention. Yet most businesses have no systematic process for re-engaging dormant contacts. They just... disappear. This workflow automates the entire win-back funnel: identify, segment, and reach out. Run it monthly to keep your inactive list from growing. Over time, you can track which win-back messages get the highest response rates and refine your approach.

Variations to try:

"Exclude contacts who have previously unsubscribed or asked to stop messages"
"Create separate groups for 30-day, 60-day, and 90-day inactive contacts with different messages for each"
"After the campaign sends, check responses after 48 hours and update lead statuses for anyone who replied"

Combining Workflows: Autonomous Operation

Each workflow above is useful on its own. The real power emerges when you chain them together into a single prompt.

Here is an example of a combined morning routine:

"Run my daily WhatsApp ops check. First, pull my lead digest for the last 24 hours. Second, find any contacted leads older than 3 days without a reply and send follow-ups. Third, check sentiment in all monitored groups and flag anything urgent. Finally, tell me my usage stats so I know how many messages I have left this month."

Claude will execute all four steps in sequence, calling 8-10 tools across the chain, and return a single consolidated report. What used to be 30 minutes of dashboard clicking becomes one prompt and a 30-second wait.

You can also break these into scheduled routines. Use Claude Code's automation features or your operating system's task scheduler to trigger prompts at specific times:

Morning (8 AM): Lead digest + sentiment check
Midday (12 PM): Follow-up sweep for stale leads
End of week (Friday 5 PM): Campaign performance review + win-back identification

The MCP tools are stateless -- every call is independent. You can run these prompts in any order, at any frequency, without worrying about conflicts or race conditions.

Beyond These Five

These workflows are starting points, not limits.

The pattern is always the same: describe the outcome you want in plain language, and let Claude figure out which combination of MoltFlow's 26 tools will get you there. You do not need to memorize tool names, learn parameter formats, or write API calls. The AI handles all of that.

Some other workflows people have built:

Lead scoring reviews: "Re-analyze all leads from this week and tell me if any scores should be adjusted based on follow-up interactions"
Group comparison: "Which of my monitored groups generates the highest-quality leads? Show me average scores by group for the last 30 days"
Usage forecasting: "At my current send rate, when will I hit my plan limit this month? Should I upgrade?"
Contact research: "Pull the full history for this phone number across all sessions -- every message, every group mention, every lead capture"

If you can describe it, you can automate it. The MCP connection between Claude and MoltFlow turns natural language into WhatsApp actions. No code, no configuration, no switching between tools.

Set up the MCP connection once, and every future automation is just a sentence away. Get started at molt.waiflow.app/mcp.