Build Your First AI Agent in 30 Minutes

Building Your First AI Agent in 30 Minutes

Most people think building AI agents requires machine learning expertise, cloud infrastructure, and weeks of development. Reality: you can deploy a working agent in one focused session, no coding required.

This guide walks through building a real lead qualification agent from scratch—one that researches prospects, scores them, and triggers outreach—in 30 minutes.

What You'll Build

A lead qualification agent that:

1. Receives new leads from a form or spreadsheet
2. Researches each company (visits website, checks LinkedIn)
3. Scores based on fit criteria (industry, size, location)
4. Routes high-score leads to immediate outreach
5. Archives low-score leads for future nurture

By the end, you'll have an autonomous system processing leads without manual qualification.

Prerequisites (5 minutes)

What You Need

Lead source: Google Sheet, Airtable, or form submissions Output destination: CRM, email, or another sheet Qualification criteria: How do you decide if a lead is qualified?

Example Criteria

For a B2B software targeting small businesses:

• ✅ Company size: 10-200 employees
• ✅ Industry: Professional services, healthcare, or manufacturing
• ✅ Location: United States
• ✅ Has website (shows they're real business)
• ❌ Individual/solopreneur
• ❌ Wrong industry
• ❌ Outside service area

Write down your criteria. The agent will apply them automatically.

Step 1: Connect Your Lead Source (5 minutes)

Option A: Google Sheets

1. Create sheet with columns: Company Name, Website, Contact Email, Location
2. Add 5-10 sample leads
3. Connect to agent platform (provide sheet URL, grant read access)

Option B: Form Submissions

1. Set up webhook from your form tool (Typeform, Gravity Forms, etc.)
2. Agent receives new submissions in real-time
3. Test with one submission

Option C: CRM Integration

Connect to HubSpot, Salesforce, or Pipedrive:

1. Authenticate via OAuth
2. Agent monitors "new lead" events
3. Triggers on any new record in specified view/stage

For this guide, we'll use Google Sheets (simplest to set up).

Step 2: Define Research Actions (5 minutes)

Tell the agent what to research about each lead:

For each company:

1. Visit their website homepage
2. Look for "About" page, extract company description
3. Count approximate team size (from "Team" or "About" page)
4. Identify industry from content and services offered
5. Check for LinkedIn company page, note follower count
6. Extract any visible location/headquarters info

Output: Structured data about each lead's business.

The agent executes this research workflow for every new lead automatically.

Step 3: Build Scoring Logic (5 minutes)

Define how to score leads based on research:

Score = 0 (start)

IF company_size >= 10 AND company_size <= 200:
  Score +30

IF industry IN ["Professional Services", "Healthcare", "Manufacturing"]:
  Score +25

IF location = "United States":
  Score +20

IF has_website = TRUE:
  Score +15

IF linkedin_followers > 500:
  Score +10

Total possible: 100 points

Thresholds:

• 70-100: Hot lead (immediate outreach)
• 40-69: Warm lead (nurture campaign)
• 0-39: Cold lead (archive)

Agent applies scoring automatically and routes based on thresholds.

Step 4: Configure Actions (5 minutes)

Define what happens at each score level:

Hot Leads (70-100)

1. Add to CRM with "High Priority" tag
2. Send immediate Slack notification to sales team
3. Trigger personalized email sequence
4. Add to weekly sales review list

Warm Leads (40-69)

1. Add to CRM with "Nurture" tag
2. Add to monthly newsletter list
3. Schedule follow-up check in 30 days

Cold Leads (0-39)

1. Log to archive sheet
2. Note disqualification reason
3. No immediate action

Agent executes appropriate workflow based on score.

Step 5: Test with Sample Data (5 minutes)

Add Test Leads

Create 5 test leads in your sheet:

1. Perfect fit: Mid-size healthcare company, US, has website
2. Good fit: Small professional services, US
3. Marginal: Right size but wrong industry
4. Bad fit: Solopreneur, no website
5. Bad fit: Large enterprise (too big)

Run Agent

Trigger manual run:

1. Agent reads 5 leads
2. Researches each (visits websites, checks LinkedIn)
3. Scores based on criteria
4. Routes to appropriate actions

Verify Results

• Lead 1 scored 85 → Added to CRM as hot, Slack notification sent ✓
• Lead 2 scored 75 → Added to CRM as hot ✓
• Lead 3 scored 55 → Added to nurture campaign ✓
• Lead 4 scored 15 → Archived ✓
• Lead 5 scored 30 → Archived ✓

If results look good, agent is working correctly.

Step 6: Activate Automation (5 minutes)

Schedule or Trigger

Choose execution mode:

Scheduled: Runs daily at 8 AM, processes new leads Real-time: Runs immediately when new lead added to sheet Manual: You trigger when ready to process batch

For active lead flow, choose real-time or daily scheduled.

Set Notifications

Agent notifies you:

• Daily summary (X leads processed, Y qualified)
• Immediate alert for hot leads (scored 70+)
• Weekly report (trends, score distribution)

No news is good news—agent handles routine work silently.

Monitor First Week

Watch closely for 5-7 days:

• Are scores accurate?
• Are hot leads actually qualified?
• Any patterns in cold leads (systematic issue?)

Adjust criteria or thresholds based on real results.

Real-World Results

After deploying a 30-minute lead qualification agent:

Before: Manual qualification took 10-15 minutes per lead, happened sporadically, many leads never reviewed.

After: 100% of leads researched and scored within 10 minutes of arrival, zero manual qualification time, hot leads routed instantly.

Impact: Sales team gets only pre-qualified leads with full context. Close rate increased 35% (better targeting). Time-to-first-contact decreased from 24 hours to 30 minutes (automated routing).

Maintenance: 15 minutes/week reviewing agent activity and score accuracy.

Common Issues and Fixes

Issue: Agent Scores Everything Low

Cause: Criteria too strict, very few leads can score high.

Fix: Lower thresholds or add points for more criteria. Aim for 15-20% of leads scoring 70+.

Issue: Agent Can't Find Company Info

Cause: Website link is broken or missing.

Fix: Add email domain extraction (research company by domain if website fails).

Issue: Research Takes Too Long

Cause: Agent visiting too many pages per lead.

Fix: Limit to homepage + one additional page. Speed matters more than exhaustive research.

Issue: Hot Leads Aren't Actually Qualified

Cause: Scoring criteria don't match your actual ICP.

Fix: Review last 10 closed deals. What do they have in common? Adjust criteria to match.

Expanding Your Agent

Once basic qualification works, add:

Week 2: Email Enrichment

Agent finds decision-maker emails:

• Checks company website contact page
• Uses LinkedIn to identify likely decision-makers
• Validates email deliverability
• Adds verified emails to CRM

Week 3: Personalized Outreach

Agent drafts custom emails:

• References specific company details from research
• Mentions relevant use case or industry challenge
• Includes clear call-to-action
• Sends from your email, tracks opens

Week 4: Follow-Up Sequences

Agent manages multi-touch campaigns:

• Day 0: Initial outreach
• Day 3: Follow-up if opened but no reply
• Day 7: Final touch with alternative angle
• Routes replies to you immediately

Month 2: Full Pipeline

Agent owns entire lead lifecycle:

• Discovery (finds new leads)
• Research (qualifies them)
• Outreach (makes first contact)
• Follow-up (nurtures to response)
• Handoff (routes to sales when ready)

All autonomous, minimal oversight.

What Makes This Different

No Code Required

You define what should happen in plain language. No Python scripts, no API docs, no deployment.

Adapts Automatically

Agent reads websites as they exist today. When a site redesigns, agent adapts. No broken scripts.

Decision-Making

Agent applies judgment ("Is this company description about healthcare?") not just rules ("Does text contain word 'healthcare'?").

End-to-End

Agent doesn't just score leads—it researches, scores, routes, and tracks. Complete workflow, not a helper tool.

Cost and Scale

30-Minute Agent Economics

Setup time: 30 minutes Monthly cost: $50-150 (depending on lead volume) Monthly time: 1 hour oversight Leads processed: 200-2,000/month (scales automatically)

vs Traditional Approaches

Manual qualification: 10-15 min/lead × 200 leads = 33-50 hours/month Virtual assistant: $1,200-2,000/month + training + management Lead scoring platform: $500-1,500/month + setup + maintenance

Agent delivers comparable results at fraction of cost and effort.

Your First Agent Checklist

• Define qualification criteria (10 minutes thinking)
• Connect lead source (Google Sheet, form, or CRM)
• Configure research workflow (what to look for)
• Build scoring logic (points for each criterion)
• Set up routing actions (what happens at each score)
• Test with 5-10 sample leads
• Verify results match expectations
• Activate automation (schedule or real-time)
• Monitor first week closely
• Adjust based on results

Next Steps

After your first agent proves value:

1. Add second agent: Maybe outreach or follow-up
2. Increase volume: Process more leads per day
3. Add data sources: Pull leads from multiple channels
4. Improve scoring: Refine based on closed deal patterns
5. Build agent team: Multiple specialists working together

Start simple. Prove value. Scale from there.

Conclusion

Building an AI agent isn't a months-long engineering project. With modern platforms, you define the goal and workflow in plain language, connect your data sources, and deploy. The 30-minute lead qualification agent described here is production-ready and handles real work autonomously.

The key is starting with a narrow, well-defined workflow (qualify leads) rather than trying to automate everything at once. Once that works, expand incrementally.

Most businesses that deploy their first agent in 30 minutes have 3-5 agents running within 90 days, handling significant portions of marketing, sales, and operations autonomously.

Ready to build your first agent? Start with Actus Agent and go from zero to autonomous qualification in 30 minutes.