Skip to content

01-first-campaign.md

Latest commit

 

History

History
253 lines (189 loc) · 5.91 KB

01-first-campaign.md

File metadata and controls

253 lines (189 loc) · 5.91 KB

Tutorial 1: Your First Campaign

Time: 15 minutes Difficulty: Beginner

Learn how to submit a campaign to the AI Marketing Swarm and watch it optimize in real-time.

Goal

By the end of this tutorial, you will:

  • Submit a campaign to the swarm
  • Monitor optimization progress
  • View recommendations from the agents

Prerequisites

  • Completed Getting Started
  • Swarm is running (npx @claude-flow/cli@latest swarm status shows active)

Step 1: Create a Campaign Object

First, let's define our campaign:

// campaign.ts
import { Campaign, Platform } from '@marketing/ai-swarms';

const campaign: Campaign = {
  id: 'summer-promo-001',
  name: 'Summer Promotion 2024',
  platform: 'google-ads',
  status: 'active',
  budget: {
    daily: 500,
    total: 15000,
  },
  targeting: {
    geoTargets: ['US', 'CA', 'UK'],
    ageRange: { min: 25, max: 54 },
    interests: ['outdoor', 'fitness', 'travel'],
  },
  bidStrategy: 'maximize_conversions',
  createdAt: new Date(),
  updatedAt: new Date(),
};

Step 2: Submit for Optimization

Now submit the campaign to the swarm:

import { startMarketingSwarm } from '@marketing/ai-swarms';

async function optimizeCampaign() {
  // Start the swarm
  const swarm = await startMarketingSwarm();

  // Submit the campaign
  await swarm.submitCampaign(campaign);

  console.log('Campaign submitted for optimization!');

  // Submit an optimization task
  const task = await swarm.submitTask('campaign_optimization', {
    campaign,
    goals: {
      primaryMetric: 'conversions',
      targetCpa: 25.00,
    },
  });

  console.log(`Task ID: ${task.id}`);
  console.log(`Status: ${task.status}`);

  return task;
}

optimizeCampaign();

Step 3: Monitor Progress

Watch the agents work:

async function monitorProgress(swarm, taskId: string) {
  // Check status every 5 seconds
  const interval = setInterval(async () => {
    const status = swarm.getStatus();
    const metrics = swarm.getMetrics();

    console.log('\n--- Swarm Status ---');
    console.log(`Active Agents: ${status.activeAgents}`);
    console.log(`Tasks Completed: ${metrics.totalTasksCompleted}`);
    console.log(`Avg Duration: ${metrics.averageTaskDuration.toFixed(0)}ms`);

  }, 5000);

  // Stop monitoring after 1 minute
  setTimeout(() => {
    clearInterval(interval);
    console.log('\nMonitoring complete!');
  }, 60000);
}

Step 4: View Recommendations

The agents will generate recommendations. Here's how to access them:

// Listen for optimization events
swarm.on('task:completed', (event) => {
  if (event.payload?.recommendations) {
    console.log('\n--- Agent Recommendations ---');
    for (const rec of event.payload.recommendations) {
      console.log(`• ${rec.type}: ${rec.description}`);
      console.log(`  Expected Impact: ${rec.expectedImpact}`);
      console.log(`  Confidence: ${(rec.confidence * 100).toFixed(0)}%`);
    }
  }
});

Example output:

--- Agent Recommendations ---
• bid_adjustment: Increase max CPC for "outdoor gear" keywords by 15%
  Expected Impact: +12% conversions
  Confidence: 78%

• audience_expansion: Add lookalike audience based on converters
  Expected Impact: +8% reach at similar CPA
  Confidence: 85%

• schedule_optimization: Shift 20% budget to 6-9 PM slots
  Expected Impact: -5% CPA
  Confidence: 72%

Complete Example

Here's the full working code:

// first-campaign.ts
import { startMarketingSwarm, Campaign } from '@marketing/ai-swarms';

async function main() {
  // 1. Define the campaign
  const campaign: Campaign = {
    id: 'summer-promo-001',
    name: 'Summer Promotion 2024',
    platform: 'google-ads',
    status: 'active',
    budget: { daily: 500, total: 15000 },
    targeting: {
      geoTargets: ['US'],
      ageRange: { min: 25, max: 54 },
    },
    bidStrategy: 'maximize_conversions',
    createdAt: new Date(),
    updatedAt: new Date(),
  };

  // 2. Start the swarm
  console.log('Starting AI Marketing Swarm...');
  const swarm = await startMarketingSwarm();
  console.log(`✓ Swarm started with ${swarm.getStatus().activeAgents} agents`);

  // 3. Submit the campaign
  console.log('\nSubmitting campaign...');
  await swarm.submitCampaign(campaign);
  console.log('✓ Campaign submitted');

  // 4. Request optimization
  console.log('\nRequesting optimization...');
  const task = await swarm.submitTask('campaign_optimization', {
    campaign,
    goals: { primaryMetric: 'conversions', targetCpa: 25.00 },
  });
  console.log(`✓ Task created: ${task.id}`);

  // 5. Wait and check results
  console.log('\nWaiting for agents to analyze...');
  await new Promise(resolve => setTimeout(resolve, 5000));

  const metrics = swarm.getMetrics();
  console.log(`\n--- Results ---`);
  console.log(`Tasks Completed: ${metrics.totalTasksCompleted}`);
  console.log(`Success Rate: ${((1 - metrics.errorRate) * 100).toFixed(1)}%`);

  // 6. Cleanup
  await swarm.stop();
  console.log('\n✓ Done!');
}

main().catch(console.error);

Run it:

npx tsx first-campaign.ts

What Happened Behind the Scenes?

When you submitted the campaign, here's what the agents did:

  1. Orchestrator received the task and routed it to relevant specialists
  2. Memory retrieved similar past campaigns for context
  3. Simulation ran 1000+ Monte Carlo scenarios
  4. Historical Memory found patterns from winning campaigns
  5. Risk Detection checked for potential issues
  6. Quality validated all recommendations before output

All in milliseconds!

Verification

Your optimization worked if you see:

  • ✓ "Campaign submitted" message
  • ✓ Task ID generated
  • ✓ Tasks Completed: 1 or more in results

Next Steps