Use Case

Screenshot API for AI Agents

Give your AI agents eyes on the web. Screenshots as structured input for LLMs, vision models, and autonomous workflows.

The Problem

AI agents can't see the web

LLMs and AI agents work with text, but the web is visual. HTML source code doesn't convey layout, broken elements, or visual context. For tasks like web research, QA, and monitoring, agents need to see what users see.

LLMs can't interpret raw HTML for visual understanding
Running headless browsers in AI pipelines is complex and slow
Browser automation tools are heavy for simple visual checks
No easy way to feed web visuals into vision model APIs

The Solution

A visual perception layer for any AI agent

One API call returns a screenshot that you can pass directly to GPT-4o, Claude, or any vision model. No browser management, no puppeteer scripts.

Step 1

Agent decides to look at a page

Your AI agent determines it needs visual context for a URL.

Step 2

Call the screenshot API

A single HTTP request returns a screenshot — no browser setup needed.

Step 3

Feed to vision model

Pass the screenshot to GPT-4o, Claude, or Gemini for visual analysis.

Code Example

Get started in minutes

A single API call is all you need.

python

import requests, base64, anthropic

# 1. Capture screenshot
screenshot = requests.get(
    "https://api.porto.dev/v1/screenshot",
    params={"url": "https://competitor.com/pricing", "width": 1280},
    headers={"Authorization": "Bearer PORTO_API_KEY"},
).content

# 2. Send to Claude for analysis
client = anthropic.Anthropic()
message = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": [
            {"type": "image", "source": {
                "type": "base64",
                "media_type": "image/png",
                "data": base64.b64encode(screenshot).decode(),
            }},
            {"type": "text", "text": "Analyze this pricing page."},
        ],
    }],
)