4-Agent CRM Automation System

A complete CRM automation built on Google Sheets, Gmail, Calendly, and OpenRouter. Four AI agents handle every stage of the pipeline — from lead intake to no-show recovery — running 24/7 for approximately $5/month.

Cost: ~$5/month — Google Sheets (free) + Gmail (free) + Calendly (free) + n8n self-hosted (free) + OpenRouter (~$5/mo). No Salesforce. No HubSpot. No expensive subscriptions.

The Problem

Business owners face an impossible choice: serve existing clients or nurture new leads. Every hour spent on follow-ups is an hour not spent on delivery. A single missed follow-up on a $10,000 deal is $10,000 lost — and most businesses miss dozens every month.

Architecture

graph LR
    A[New Lead] --> B[CRM Agent]
    B --> C[Follow-up Agent]
    C --> D{Booked?}
    D -->|Yes| E[Concierge Agent]
    D -->|No| C
    E --> F{Showed up?}
    F -->|Yes| G[Sales Call]
    F -->|No| H[No-Show Agent]
    H --> C

Each agent is a separate n8n workflow that reads from and writes to the same Google Sheet — your single source of truth. No lead falls through the cracks because every agent knows exactly where every prospect stands.

The Four Agents

1

Agent 1: CRM Agent — The Foundation

What it does: Processes every new lead automatically. When someone fills out a form, sends an email, or gets added manually, this agent creates a structured entry in your Google Sheets CRM — contact details, source, timestamp, status, and any context available.

Why it matters: Without a clean, centralized record, everything else breaks. This agent ensures that from the moment a lead exists, the system knows about it.

CRM columns:

ColumnPurpose
NameLead’s full name
EmailContact email
CompanyCompany name
SourceWhere the lead came from
StatusNew / Contacted / Booked / Closed / Lost
NotesAI-generated context
Last ContactDate of most recent touchpoint
Next ActionWhat happens next and when
2

Agent 2: Follow-up Agent — The Closer

What it does: Sends 3 personalized follow-up emails at timed intervals. Each email is AI-generated based on the lead’s specific data — not a mail-merge template, but actual personalized content that references their industry, company size, and stated pain points.

The sequence:

EmailTimingContent Strategy
1ImmediateValue-first introduction. Lead with insight, not a pitch.
2+2 daysShare a relevant case study or industry insight.
3+5 daysFinal direct ask. Clear call-to-action with urgency.

How AI personalization works: The AI reads the lead’s CRM data (company name, industry, pain point) and generates each email to sound like it was written by a human who did 10 minutes of research. At scale, this is impossible manually. With AI, every lead gets the same quality of attention whether you have 10 or 1,000.

3

Agent 3: Concierge Agent — The Welcomer

What it does: Monitors Calendly for new bookings. When a lead books a call, it instantly:

  1. Updates the CRM status to “Booked”
  2. Sends a confirmation email with prep materials
  3. Notifies the salesperson with full lead context

Why it matters: The gap between “booked” and “attended” is where many deals die. This agent fills that gap with professionalism — the lead feels expected and valued.

4

Agent 4: No-Show Agent — The Recovery System

What it does: Detects when a lead books but doesn’t show. Sends a re-engagement email (“I noticed we missed each other — want to reschedule?”) and resets the follow-up sequence.

Reality check: No-show rates for sales calls range from 20-40%. Without this agent, those leads are lost. With it, a significant portion rebook and eventually convert.

Implementation Guide

1

Step 1: Set Up the CRM (15 minutes)

Create a Google Sheet with the columns from Agent 1 above. This is your CRM — it’s not Salesforce, and it doesn’t need to be. It tracks every lead, shows your pipeline at a glance, and costs nothing.
2

Step 2: Build the Follow-up Agent (1 hour)

In n8n, create a workflow:

  1. Trigger: New row in Google Sheet (or webhook from your web form)
  2. AI Node: OpenRouter generates a personalized email based on the lead’s data
  3. Gmail Node: Sends the email
  4. Update Sheet: Marks status as “Contacted” and logs the timestamp
  5. Wait Node: 2-day delay
  6. Repeat: Send email 2 (case study), wait 3 days, send email 3 (direct ask)
3

Step 3: Add Calendly Integration (30 minutes)

Create a second workflow:

  1. Trigger: Calendly webhook fires when someone books
  2. Find lead in Sheet: Match by email address
  3. Update status: Change to “Booked”
  4. Send notification: Slack or email with full lead context
4

Step 4: Run and Iterate

Feed 10 test leads into the system. Watch the automation execute. Adjust email tone, timing, and content based on what you observe. Scale when confident.

AI Nurturing Sequences

For leads not yet ready to buy, the system runs AI-personalized nurture sequences:

WeekEmail TypeAI Personalization
1Value insightIndustry-specific statistic or trend
2Case studyMatched to lead’s company size and industry
3Tool or resourceFree template relevant to their pain point
4Soft ask“Would a 15-minute call make sense?”
6Re-engagementNew angle based on recent industry news
8Final value“Here’s what companies like yours achieved”
AspectTraditional DripAI Nurturing
ContentSame email to everyonePersonalized per lead’s context
ScheduleFixed intervalsAdaptive based on engagement
LearningNo adjustmentAdjusts on opens, clicks, replies
QualityGeneric templatesIndustry-matched content

The system watches for engagement signals:

  • Lead opens email within an hour? Move up the next email — they’re interested.
  • No opens in 2 weeks? Pause, then re-engage with a different angle.
  • Lead clicks a pricing link? Alert the salesperson immediately — hot lead.

This uses n8n’s wait nodes combined with webhook triggers from your email tool.

Proposal Generator

After a sales call, AI turns transcripts into proposals in minutes, not hours:

graph LR
    A[Sales Call] --> B[Fathom.ai Records]
    B --> C[Transcript Cleanup]
    C --> D[AI Extracts 18+ Fields]
    D --> E[Populates Google Slides]
    E --> F[Human Reviews 2-5 min]
    F --> G[Send to Client]
MetricValue
Proposal creation time2-4 hours
Proposals per week3-5
ConsistencyVariable quality
Cost per proposalYour hourly rate
MetricValue
Proposal creation time2-5 minutes + review
Proposals per week15-20
ConsistencyTemplate-enforced
Cost per proposal~$0.02 per AI extraction
Human in the Loop: Never send AI-generated proposals directly to clients. The AI extracts and structures — you review and approve. A 5-minute review for a $10,000+ deal is the minimum quality gate.

Results

MetricBeforeAfter
Lead response timeHours to daysImmediate (24/7)
Forgotten follow-upsFrequentZero
ScalabilityLimited by headcountHandles 10 or 1,000 identically
Pipeline visibilityManual trackingReal-time CRM status
Proposal turnaround2-4 hours2-5 minutes
No-show recoveryLost leadsAutomated re-engagement

Cost Breakdown

ComponentCost
Google SheetsFree
GmailFree
CalendlyFree
n8n (self-hosted)Free
OpenRouter (AI)~$5/mo
Total~$5/month