Build an AI Presentation Agent: Complete Developer Guide

AI agents are transforming how software creates content. An AI presentation agent can generate, customize, and deliver professional slide decks as part of automated workflows — from CRM pipelines to internal reporting tools.

This guide walks you through building a production-ready presentation agent using the 2Slides API.

Architecture Overview

A presentation agent typically follows this flow:

User Request → Agent Logic → 2Slides API → Poll Job → Deliver Result
                  ↓
           Context Gathering
           (data, templates, brand assets)

Core Components

1. Input Handler — receives presentation requests (text, data, files)
2. Context Builder — enriches requests with templates, brand info, data
3. Generation Engine — calls 2Slides API to create slides
4. Status Monitor — polls job status until completion
5. Delivery Layer — returns download links, sends via email, posts to Slack

Getting Started

1. Get Your API Key

Sign up at 2slides.com and create an API key at 2slides.com/api.

2. Install Dependencies

npm install node-fetch dotenv

3. Basic Agent Implementation

import fetch from 'node-fetch';

class PresentationAgent {
  constructor(apiKey) {
    this.apiKey = apiKey;
    this.baseUrl = 'https://2slides.com/api/v1';
  }

  async generateSlides({ topic, themeId, language = 'en', resolution = '2K' }) {
    // Step 1: Start generation
    const response = await fetch(`${this.baseUrl}/slides/generate`, {
      method: 'POST',
      headers: {
        'Authorization': `Bearer ${this.apiKey}`,
        'Content-Type': 'application/json'
      },
      body: JSON.stringify({
        userInput: topic,
        themeId,
        responseLanguage: language,
        resolution,
        mode: 'async'
      })
    });

    const job = await response.json();
    if (!response.ok) throw new Error(job.error || 'Generation failed');

    // Step 2: Poll until complete
    return await this.waitForCompletion(job.jobId);
  }

  async waitForCompletion(jobId, maxWait = 300000) {
    const start = Date.now();
    while (Date.now() - start < maxWait) {
      const response = await fetch(`${this.baseUrl}/jobs/${jobId}`, {
        headers: { 'Authorization': `Bearer ${this.apiKey}` }
      });
      const status = await response.json();

      if (status.status === 'success') return status;
      if (status.status === 'failed') throw new Error('Generation failed');

      await new Promise(r => setTimeout(r, 3000));
    }
    throw new Error('Generation timed out');
  }

  async generateFromFile({ fileUrl, prompt, themeId }) {
    const response = await fetch(`${this.baseUrl}/slides/create-pdf-slides`, {
      method: 'POST',
      headers: {
        'Authorization': `Bearer ${this.apiKey}`,
        'Content-Type': 'application/json'
      },
      body: JSON.stringify({
        fileUrl,
        userInput: prompt,
        themeId,
        mode: 'async'
      })
    });

    const job = await response.json();
    return await this.waitForCompletion(job.jobId);
  }

  async generateWithDesign({ topic, referenceImageUrl, resolution = '2K' }) {
    const response = await fetch(`${this.baseUrl}/slides/create-like-this`, {
      method: 'POST',
      headers: {
        'Authorization': `Bearer ${this.apiKey}`,
        'Content-Type': 'application/json'
      },
      body: JSON.stringify({
        userInput: topic,
        designStyle: {
          global: { referenceImageUrl }
        },
        resolution,
        mode: 'async'
      })
    });

    const job = await response.json();
    return await this.waitForCompletion(job.jobId);
  }

  async addNarration({ jobId, mode = 'single', voice = 'Charon' }) {
    const response = await fetch(`${this.baseUrl}/slides/generate-narration`, {
      method: 'POST',
      headers: {
        'Authorization': `Bearer ${this.apiKey}`,
        'Content-Type': 'application/json'
      },
      body: JSON.stringify({ jobId, mode, voice, contentMode: 'concise' })
    });

    return await response.json();
  }
}

// Usage
const agent = new PresentationAgent('sk-2slides-xxx');

const result = await agent.generateSlides({
  topic: 'Q1 2026 Business Review',
  themeId: 'mckinsey-theme-id',
  resolution: '2K'
});

console.log('Download:', result.downloadUrl);

Python Implementation

import requests
import time

class PresentationAgent:
    def __init__(self, api_key):
        self.api_key = api_key
        self.base_url = 'https://2slides.com/api/v1'
        self.headers = {
            'Authorization': f'Bearer {api_key}',
            'Content-Type': 'application/json'
        }

    def generate_slides(self, topic, theme_id=None, language='en', resolution='2K'):
        response = requests.post(
            f'{self.base_url}/slides/generate',
            headers=self.headers,
            json={
                'userInput': topic,
                'themeId': theme_id,
                'responseLanguage': language,
                'resolution': resolution,
                'mode': 'async'
            }
        )
        response.raise_for_status()
        job = response.json()
        return self._wait_for_completion(job['jobId'])

    def generate_from_file(self, file_url, prompt, theme_id=None):
        response = requests.post(
            f'{self.base_url}/slides/create-pdf-slides',
            headers=self.headers,
            json={
                'fileUrl': file_url,
                'userInput': prompt,
                'themeId': theme_id,
                'mode': 'async'
            }
        )
        response.raise_for_status()
        job = response.json()
        return self._wait_for_completion(job['jobId'])

    def _wait_for_completion(self, job_id, max_wait=300):
        start = time.time()
        while time.time() - start < max_wait:
            response = requests.get(
                f'{self.base_url}/jobs/{job_id}',
                headers=self.headers
            )
            status = response.json()
            if status['status'] == 'success':
                return status
            if status['status'] == 'failed':
                raise Exception('Generation failed')
            time.sleep(3)
        raise TimeoutError('Generation timed out')

# Usage
agent = PresentationAgent('sk-2slides-xxx')
result = agent.generate_slides('AI Trends 2026', resolution='4K')
print(f"Download: {result['downloadUrl']}")

Advanced Patterns

Pattern 1: Template-Based Generation

Pre-configure themes for different use cases:

const TEMPLATES = {
  'pitch-deck': { themeId: 'apple-id', resolution: '2K' },
  'quarterly-review': { themeId: 'mckinsey-id', resolution: '2K' },
  'training': { themeId: 'corporate-id', resolution: '1K' },
  'social-media': { themeId: 'saul-bass-id', resolution: '4K', aspectRatio: '1:1' },
};

async function generateByType(type, topic) {
  const config = TEMPLATES[type];
  return agent.generateSlides({ topic, ...config });
}

Pattern 2: Data Pipeline Integration

Connect to your data sources:

async function weeklyReport(dataSource) {
  // 1. Fetch latest data
  const data = await dataSource.getWeeklyMetrics();

  // 2. Format as presentation input
  const topic = `Weekly Report: Revenue $${data.revenue}, Users ${data.users}, Churn ${data.churn}%`;

  // 3. Generate slides
  const result = await agent.generateSlides({ topic, themeId: 'mckinsey-id' });

  // 4. Deliver
  await slackBot.postMessage('#team-updates', `Weekly report ready: ${result.downloadUrl}`);
}

Pattern 3: Multi-Format Output

Generate slides + narration + video in one pipeline:

async function fullPresentation(topic) {
  // 1. Generate slides
  const slides = await agent.generateSlides({ topic });

  // 2. Add voice narration
  await agent.addNarration({ jobId: slides.jobId, mode: 'multi', voice: 'Charon' });

  // 3. Download with audio
  const download = await fetch(`${agent.baseUrl}/slides/download-slides-pages-voices`, {
    method: 'POST',
    headers: agent.headers,
    body: JSON.stringify({ jobId: slides.jobId })
  });

  return download.json();
}

Production Considerations

Rate Limiting

• Default: 60 requests/minute
• Handle 429 responses with exponential backoff
• Use async mode for batch operations

Error Handling

• Wrap all API calls in try/catch
• Implement retry logic for transient failures
• Log job IDs for debugging

Cost Management

• Track credit usage per generation
• Set budget alerts in your 2Slides dashboard
• Use lower resolutions for internal/draft content

Security

• Store API keys in environment variables, never in code
• Use server-side API calls only — never expose keys to clients
• Implement request signing for webhook callbacks

Frequently Asked Questions

Can I build a SaaS product on top of 2Slides API?

Yes — the API is designed for production use. There are no restrictions on commercial applications.

What's the maximum concurrent generation limit?

The API handles concurrent requests via rate limiting. For high-volume needs, contact the 2Slides team for enterprise limits.

Is there a webhook for job completion?

Currently, use polling via the jobs endpoint. Webhook support is on the roadmap.

---

Build your presentation agent — get your API key at 2Slides and start generating slides programmatically.