Intuned - AI-Powered Browser Automation as Code

TL;DR

TL;DR: Intuned is a YC-backed platform where an AI agent builds and maintains browser automations as real Playwright code, auto-healing them when target websites change.

Source and Accuracy Notes

This post is based on the Launch HN post from June 8, 2026, and the official Intuned documentation. Intuned is a Y Combinator S22 company founded by Faisal and Ahmad.

What Is Intuned?

Intuned is a managed platform for building, deploying, and maintaining browser automations. Instead of writing fragile Playwright or Puppeteer scripts that break every time a website updates its DOM, Intuned uses an AI agent to generate automation code and then continuously monitors and repairs it when sites change.

The core insight: the hardest part of browser automation is not writing the initial script, it is keeping it working over time. Selectors break, layouts shift, and debugging failures across isolated environments is painful. Intuned addresses this by treating automations as managed code that an AI agent owns end-to-end.

How It Works

Infrastructure Layer

Every project runs in its own isolated machine on Intuned’s infrastructure. The platform handles:

• Auth and session reuse across runs
• Scheduling and batch execution
• Concurrency management
• Observability and run context capture
• CAPTCHA solving helpers

Projects are typically Playwright-based TypeScript or Python. You can write them directly in Intuned’s online IDE or let the AI agent generate them from a natural language description.

AI Agent Layer

The agent builds on the Claude Agent SDK. Rather than following a rigid pipeline (collect requirements, inspect site, generate code, patch breaks), it operates more like a coding agent that can reason about the current state of a website and adapt.

When a website changes and an automation breaks, the agent:

1. Receives the failure context (DOM snapshot, error trace, run logs)
2. Analyzes what changed on the target page
3. Generates a fix to the underlying Playwright code
4. Deploys the patched version automatically

This is fundamentally different from selector-based RPA tools that try to match elements by position or CSS path. Intuned’s agent understands the semantic intent of each step and can find alternative ways to accomplish the same goal when the UI changes.

Web Tasks

Intuned also offers a higher-level abstraction called Web Tasks, where you describe a repeatable web operation in natural language and the platform executes it using a mix of AI and generated code. This is useful for operations that do not need full code-level control.

Setup Workflow

Step 1: Create an Account and Project

Sign up at intunedhq.com and create a new project. Choose between TypeScript or Python as your runtime language.

Step 2: Describe Your Automation

Use the AI agent to describe what you want to automate in plain English. For example:

Navigate to the dashboard, log in with stored credentials,
download the monthly report as PDF, and upload it to S3.

The agent inspects the target website and generates Playwright code that accomplishes the task.

Step 3: Deploy and Schedule

Once the automation is generated, deploy it to Intuned’s runtime. Set up a schedule (cron, webhook, or manual trigger). The platform handles isolated execution, session management, and output delivery.

Step 4: Monitor and Auto-Heal

When a run fails, Intuned captures the full context (DOM, screenshots, network logs). The AI agent reviews the failure and proposes or applies a fix. You get notified of changes and can review patches before they go live.

SDK and CLI

Intuned provides SDKs in both Python and TypeScript for programmatic control:

from intuned import IntunedClient

client = IntunedClient(api_key="your-key")

# Run a web task
result = client.web_tasks.run(
    task_id="monthly-report",
    params={"month": "2026-06"}
)

import { IntunedClient } from "@intuned/sdk";

const client = new IntunedClient({ apiKey: "your-key" });

const result = await client.webTasks.run({
  taskId: "monthly-report",
  params: { month: "2026-06" },
});

The CLI supports project management, local testing, and deployment:

# Initialize a new project
intuned init

# Run locally for testing
intuned run

# Deploy to production
intuned deploy

Deeper Analysis

Why Code-First Matters

Many browser automation tools operate as black-box recorders or visual editors. They produce automations that are opaque and hard to debug. Intuned’s code-first approach means every automation is real Playwright code that you can read, version-control, and reason about. If the AI agent’s fix does not look right, you can override it manually.

The Maintenance Problem

Industry estimates suggest that 60 to 80 percent of the cost of browser automation is in maintenance, not initial development. Websites change constantly, and each change can break selectors, timing assumptions, or navigation flows. Intuned’s auto-heal capability directly targets this cost center.

When to Use Intuned

Good fits:

• Scraping data from websites without APIs
• Automating form submissions and report downloads
• Pulling data from internal tools that lack REST APIs
• E-commerce price monitoring across changing layouts

Less ideal:

• High-frequency trading or sub-second latency requirements
• Automations that need to run on your own infrastructure for compliance reasons
• Simple one-off scripts where the maintenance overhead is not justified

Practical Evaluation Checklist

• [ ] Does the AI agent generate working Playwright code on the first try for your target site?
• [ ] How quickly does auto-heal respond when a selector breaks?
• [ ] Can you review and approve agent-generated patches before deployment?
• [ ] Does the isolated runtime handle your auth flow (SSO, 2FA, session cookies)?
• [ ] Are run logs and screenshots detailed enough to debug failures manually?
• [ ] Does the pricing model fit your execution volume?

Security Notes

• Automations run in isolated machines, not shared containers
• Credentials and session data are managed by the platform’s auth layer
• Review the privacy policy for data handling details
• Consider whether target websites’ terms of service permit automated access
• API keys should be stored in Intuned’s secrets management, not in project code

FAQ

Q: How is Intuned different from traditional RPA tools like UiPath? A: Intuned generates real Playwright code rather than proprietary automation scripts. The AI agent maintains the code when sites change, whereas traditional RPA requires manual selector updates. Intuned is also API-first and designed for developers, not business analysts.

Q: What happens when the AI agent’s auto-heal makes a wrong fix? A: You can configure the platform to require human approval before deploying patches. The full run context (DOM, screenshots, logs) is preserved so you can debug manually if needed.

Q: Can I use Intuned for sites that require login? A: Yes. Intuned handles auth and session reuse across runs. You can store credentials securely in the platform and the runtime manages session persistence.

Q: What languages are supported? A: Python and TypeScript, both using Playwright as the underlying browser automation framework.

Q: Is there a free tier or trial? A: Check the plans and billing page for current pricing. Intuned offers tiered plans based on execution volume.

Conclusion

Intuned tackles the maintenance problem that makes browser automation expensive in practice. By combining a managed Playwright runtime with an AI agent that owns the code lifecycle, it offers a compelling alternative to both hand-written scripts and black-box RPA tools. The code-first approach means you never lose visibility into what your automations are doing, and the auto-heal capability directly addresses the biggest cost driver in production browser automation.

If you are running browser automations at scale and spending more time fixing broken selectors than building new ones, Intuned is worth evaluating.