This tutorial walks you through building a complete AI-powered sales outreach sequence with Lindy AI, from prospect research to multi-touch follow-up and lead scoring.

Step 1: Define Your Outreach Strategy and Target Profile

Before building, clarify your outreach parameters:

• Target persona: “VP of Engineering or CTO at B2B SaaS companies with 50-200 employees, currently using Jira or Asana but expressing dissatisfaction with project visibility.”
• Sequence structure: 4-touch sequence — initial email, 3-day follow-up, 7-day value-add email, 14-day final check-in.
• Key messaging themes: Sprint visibility, stakeholder reporting, cross-team coordination, onboarding speed.
• Success signals: Email opens, link clicks, reply (positive or neutral), meeting booking.
• Escalation trigger: Any reply or meeting booking → immediately hand to human sales rep for closing.

Document these parameters in a brief — the Lindy uses this as its strategic context.

Step 2: Create the Outreach Lindy with Research Capability

In Lindy AI, create a new Lindy with this configuration:

“You are a sales outreach agent for TaskFlow, a project management platform. Your job is to research prospects, draft personalized outreach emails, schedule follow-ups, and hand off engaged leads to a human sales rep.

When a new prospect is added to the ‘Outreach Queue’ in HubSpot:

1. Research the prospect: Check their LinkedIn profile (via LinkedIn integration), their company website, and any recent news mentions. Identify their role, team size, current tools, recent challenges, and personal interests.
2. Draft a personalized initial email: Reference a specific challenge they likely face (based on their role and company), mention a relevant TaskFlow feature that addresses it, and include one personal touch drawn from your research (e.g., a recent company announcement or a shared industry event).
3. Send the email via Gmail, logged in HubSpot.
4. Monitor for engagement: Track opens, clicks, and replies.
5. If no reply after 3 days, send follow-up email #2 referencing the initial email and adding a new value point.
6. If no reply after 7 days, send value-add email #3 with a relevant case study or data point.
7. If no reply after 14 days, send final check-in offering a brief exploratory call.
8. If any reply is received (positive, neutral, or curious), immediately escalate to the assigned sales rep with a full context summary including all research notes, emails sent, and engagement data.”

This instruction set gives the Lindy a complete outreach playbook — research, personalize, execute, follow up, and escalate — all defined in natural language.

Step 3: Connect HubSpot and Gmail as Triggers and Actions

Set up the integrations:

• Trigger: “When a new contact is added to the ‘Outreach Queue’ pipeline stage in HubSpot, activate this Lindy.”
• Action — Research: Lindy accesses LinkedIn and web data via built-in research integrations.
• Action — Draft & Send: Lindy drafts emails and sends them through the connected Gmail account.
• Action — Log: Lindy logs all sent emails, engagement events, and status updates back into HubSpot contact records.

These connections create a closed loop: prospect enters HubSpot → Lindy researches → drafts email → sends via Gmail → logs back in HubSpot → monitors engagement → triggers follow-ups or escalation → human rep receives context packet in Slack.

Step 4: Build Email Templates with AI Personalization Instructions

Create base email templates in Lindy’s workspace that include personalization instructions:

Initial Email Template:

Subject: [AI generates subject line referencing prospect’s specific challenge]

Hi [first_name],

[AI inserts 1-2 sentences referencing the prospect’s specific situation — their role, their company’s recent activity, or a challenge they likely face based on research]

I noticed [company_name] is [AI inserts research-based observation about their project management context]. Teams like yours often struggle with [specific challenge — sprint visibility / stakeholder reporting / cross-team coordination]. TaskFlow was built to solve exactly this — [AI inserts relevant feature explanation].

[AI adds one personal touch from research — reference to a recent company milestone, shared conference attendance, or mutual industry challenge]

Would you be open to a 15-minute call this week to explore whether TaskFlow could help [company_name] deliver projects more smoothly?

Best, [Sender name]

The bracketed AI instructions tell the Lindy what to generate from its research — the template provides structure while AI fills authentic personalization.

Step 5: Configure Engagement Tracking and Follow-Up Logic

Lindy AI monitors email engagement through connected email tracking:

• “Check for email opens within 48 hours. If opened but no reply, adjust follow-up email #2 to reference the specific content they likely read.”
• “Check for link clicks. If the prospect clicked the case study link, follow-up #3 should expand on that case study’s results.”
• “If a reply contains scheduling language (‘let’s chat,’ ‘call me,’ ‘available Tuesday’), immediately escalate with priority=’Hot Lead.'”
• “If a reply is negative (‘not interested,’ ‘no budget’), log as ‘Closed — Not Interested’ in HubSpot and stop the sequence.”

This conditional logic makes the sequence adaptive — every follow-up responds to the prospect’s actual behavior rather than blindly sending static template emails on a timer.

Step 6: Set Up Lead Scoring and Handoff Protocol

Configure the escalation handoff for maximum rep efficiency:

• “When escalating a lead, send a Slack message to the assigned sales rep containing: Prospect name, company, role, all research notes, email sequence sent (with timestamps), engagement signals observed, and suggested talking points for the call.”
• “Mark the HubSpot contact as ‘Hot Lead — Awaiting Rep Contact’ and assign to the correct rep based on territory.”

The human rep receives a complete dossier — no research needed, no context gathering, no “who is this person?” confusion. They step directly into a informed conversation, increasing conversion rates on warm leads.

Step 7: Scale and Optimize the Outreach Engine

As the system runs, optimize based on performance data:

• Monitor open rates across Lindy-drafted emails vs. your previous template-based sequences. AI-personalized emails typically see 2-3x higher open rates.
• Track reply rates by personalization type — which research signals (role-based, company-event-based, personal-interest-based) drive the most responses? Double down on the most effective signals.
• Measure rep conversion rate on Lindy-handoffs vs. self-researched leads — reps closing Lindy-prepared leads should convert faster due to better context.
• Adjust sequence timing based on engagement patterns — if most opens happen within 4 hours, shorten the first follow-up to 2 days instead of 3.

Use these insights to refine the Lindy’s instructions monthly, making each generation of outreach smarter than the last.

Pro Tips for Lindy AI Sales Outreach

• Start with 20-30 prospects to test the Lindy’s research and personalization quality — review every drafted email manually before enabling auto-send.
• Add a “Do Not Contact” list in HubSpot that the Lindy checks before drafting — prevent outreach to existing customers, competitors, or opt-out contacts.
• Use separate Lindies for different persona segments — one optimized for CTO outreach, another for VP of Engineering, each with role-specific research instructions and messaging themes.
• Schedule a weekly “Outflow Report” Lindy that summarizes: prospects contacted, emails sent, engagement rates, escalations, and pipeline impact — giving you a complete outbound performance dashboard without manual reporting.

Lindy AI turns sales outreach from a labor-intensive, template-driven grind into an intelligent, research-backed, adaptive engine. By automating prospect research, personalized drafting, multi-touch sequencing, engagement tracking, and informed handoff, your sales team focuses entirely on closing conversations with pre-qualified, context-rich leads — while the AI handles everything upstream with authenticity that prospects can’t distinguish from human effort.

This tutorial shows you how to build a complete AI-powered customer support workflow with Lindy AI, from agent creation to escalation logic and performance tracking.

Step 1: Map Your Support Workflow Before Building

Before configuring Lindy AI, document your current support process:

• Channels: Where do support requests arrive? (Email, Slack, Intercom, Twitter DMs, Help desk tickets)
• Triage categories: What are your 5-8 most common request types? (Account issues, billing questions, feature requests, bug reports, integration help, onboarding, general inquiries)
• Response templates: Which requests have standard answers? (FAQ-level queries, account reset procedures, pricing explanations)
• Escalation triggers: When must a human intervene? (Security concerns, billing disputes, custom integration requests, angry customers, ambiguous technical issues)

This mapping exercise gives you the structure to configure Lindies precisely — each agent handles a specific slice of the workflow with clear boundaries.

Step 2: Create Your First Support Lindy

In Lindy AI’s dashboard, click “Create New Lindy” and configure it with natural language:

“You are a customer support agent for TaskFlow, a project management SaaS product. When a new support email or Slack message arrives, read it carefully and classify it as one of these categories: Account Issue, Billing Question, Bug Report, Feature Request, Integration Help, Onboarding, or General Inquiry. For Account Issues and Billing Questions, draft a response using our standard templates (see knowledge base). For Bug Reports and Integration Help, collect all relevant details (error messages, steps to reproduce, browser/device info) and create a structured bug report. For General Inquiries and Feature Requests, draft a friendly acknowledgment and route to the product team. If you’re unsure about the correct response or the customer seems frustrated, escalate to a human agent with a full context summary.”

This single instruction block defines the Lindy’s entire behavior — classification, drafting, data collection, routing, and escalation — all in plain English.

Step 3: Connect Support Channels as Triggers

Link your support channels so the Lindy activates automatically:

• Gmail Trigger: “When a new email arrives in support@taskflow.com, activate this Lindy.”
• Slack Trigger: “When a new message appears in the #support channel, activate this Lindy.”
• Intercom Trigger: “When a new conversation is created in Intercom, activate this Lindy.”

Lindy AI connects to these apps via built-in integrations — no API coding or webhook configuration needed. Each trigger fires independently, so the Lindy handles requests from every channel simultaneously.

Step 4: Build the Knowledge Base for Accurate Responses

Create a Notion or Google Docs page as your “Support Knowledge Base” and connect it to the Lindy:

• Document your standard response templates for each triage category.
• Include FAQ answers, account reset procedures, pricing details, and feature comparison tables.
• Add troubleshooting guides for common bugs and integration setup instructions.

In the Lindy configuration, add: “Reference the Support Knowledge Base in Notion for drafting responses. Always use information from the knowledge base rather than generating answers from general knowledge.”

This ensures the Lindy’s responses are factually grounded in your actual product documentation — not generic AI guesses.

Step 5: Configure Escalation Logic with Confidence Thresholds

Lindy AI’s smartest feature is its confidence-based escalation. Set thresholds:

• “If your confidence in the correct response is below 70%, escalate to a human.”
• “If the customer uses words like ‘frustrated,’ ‘angry,’ ‘unacceptable,’ or ‘refund,’ immediately escalate regardless of confidence.”
• “If the request involves billing disputes, security concerns, or data deletion, always escalate to a human agent.”

The Lindy evaluates each incoming request, calculates its confidence, and either drafts a response or escalates with a full context packet:

“Escalation Report: Customer Sarah Chen (sarah@company.com) submitted a billing dispute. She was charged $$49.99 instead of the advertised$$29.99 promotional rate. Confidence in auto-response: 35%. Category: Billing Dispute. Suggested action: Route to billing specialist for manual review. Full email text attached.”

This context packet gives the human agent everything needed to resolve the issue immediately — no re-reading, no re-classification, no wasted time.

Step 6: Set Up Response Quality Review

Initially, review every Lindy-drafted response before it’s sent. In the Lindy configuration:

• “Before sending any drafted response, hold it for human review. Send the draft to the #support-review Slack channel for approval.”

As you gain confidence in the Lindy’s quality, relax the review requirement:

• After 50 successfully reviewed responses with <5% edit rate, change to: “Send responses automatically for Account Issues, Billing Questions, and General Inquiries. Hold Bug Reports and Integration Help for review.”
• After 200 successful auto-responses, enable full automation with weekly quality audits.

This graduated approach prevents embarrassing AI errors while building trust in the system’s accuracy.

Step 7: Track Performance and Optimize

Monitor your support Lindy’s performance weekly:

• Volume handled: How many requests did the Lindy process vs. total incoming?
• Auto-resolved rate: Percentage of requests resolved without human intervention
• Escalation rate: Percentage escalated to humans — track whether this decreases over time as the knowledge base improves
• Customer satisfaction: Compare CSAT scores for Lindy-handled vs. human-handled requests
• Average response time: Lindy responses arrive in seconds vs. hours for human agents

Use these metrics to identify weak spots. If the Lindy struggles with Integration Help, expand that section in the knowledge base. If escalation rate stays high for Billing Questions, add more detailed billing templates.

Pro Tips for Lindy AI Support Automation

• Start with your highest-volume, lowest-complexity categories — FAQ-level requests where the Lindy can prove its accuracy quickly before tackling complex bug reports.
• Update the knowledge base every time a human agent resolves a novel issue — the Lindy learns from new documentation, not from past interactions alone.
• Use multiple Lindies for different channels — one optimized for email (longer, more detailed responses) and another for Slack/Intercom (shorter, conversational responses).
• Set a daily summary trigger: “At 6 PM, generate a summary of all support requests handled today, including categories, resolution rates, and escalated cases.” This gives you a daily support report without manual logging.

Lindy AI transforms customer support from a reactive, human-dependent operation into a proactive, AI-accelerated workflow. By combining intelligent classification, grounded responses, smart escalation, and graduated automation, you build a support system that resolves common issues in seconds while preserving human judgment for the cases that genuinely need it — delivering both speed and quality simultaneously.