Perform web actions in a cloud browser
Let your agent search, click, and fill forms to gather the information it needs. This demo extracts search results with titles, links, and summaries.
Cloud browser API for AI agentsAutomate real web tasks
Cloud Browser API provides isolated browser sessions on demand for browsing, clicking, and filling forms. Web Fetch API extracts titles, text, and links from public pages in one request. No browsers to deploy or maintain.
Web interaction and extraction
Use Cloud Browser to control cloud browsers for searches, clicks, and other interactions; use Web Fetch to extract public web content.
Task templates and tutorials
Adapt working examples to quickly validate your use case.
Keep web automation moving past CAPTCHAs
Enable automatic CAPTCHA handling through the Lexmount SDK to keep web tasks moving when verification appears. Lexmount Cloud Browser completes verification and continues subsequent operations in the same session, without human takeover or restarting the task.
Your intranet, now in a cloud browser
Open your company intranet and VPN sites in a cloud browser. No public IP, port forwarding, or publishing internal services to the internet.
Cloud Browser · Moli runtime
Create a hosted Moli Session in the Lexmount console and open a public webpage in the live browser view.
Browser engine selection
When creating a Cloud Browser session, choose Chromium or Moli based on your task’s compatibility, visual fidelity, and resource needs.
When unsure which runtime fits, choose Chromium
- Browser behavior and visual fidelity
- Persistent Context
- Extensions / downloads / recording / Replay
When your task is structure-first, choose Moli
- DOM / text / links
- Forms and network responses
- Layout and screenshots on demand
Why Moli is lighter
Moli keeps the page structure and interaction state in memory, rendering only when a visual view or screenshot is needed. In observations of internal RL training workloads, Moli used about one-seventh of Chromium’s idle memory and one-third of its peak memory, with roughly five times the average cold-start speed.
Run diagnostics and review
Use live views, available replays, and extraction results to diagnose issues and review task execution.
Find a session
Search by session ID or status.
Inspect and act
Inspect live, take over, or stop.
Review results
Review Replay when available.
Connect your agents and apps
Connect your agent to a cloud browser with Skills or MCP, or access cloud resources from your app through the SDK or API.
Connect Cloud Browser in Codex
Install the Browser Skill, authorize, and run your first web task.
Get Started
Stop maintaining browsers without losing control of the run
Checking your account. Please wait.