Browser Automation

Komo’s Cloud Browser Automation enables agents to interact with websites, web applications, and online services through real web browsers running in secure cloud infrastructure. Agents can navigate sites, click buttons, fill forms, extract data, and complete complex workflows—all without accessing your local browser or computer.

​

What is Cloud Browser Automation?

Cloud Browser Automation provides each agent with access to real Chrome browsers running in Komo’s cloud infrastructure. Unlike API integrations which are limited to what services expose, cloud browsers can interact with any website like a human would:

• Navigate websites - Visit URLs, follow links, search, browse
• Interact with web apps - Click buttons, fill forms, upload files
• Handle authentication - Log into accounts with saved sessions
• Extract data - Scrape content, download files, capture screenshots
• Execute JavaScript - Interact with dynamic single-page applications
• Complete workflows - Multi-step processes across multiple sites

All browser activity happens in the cloud—never accessing your local browser, history, or credentials.

​

Why Cloud Browser vs. Local Browser

​

The Local Browser Problem

Traditional browser automation uses your local browser, which creates security and operational risks: Security Risks:

• ❌ Access to your personal browsing history
• ❌ Access to saved passwords and cookies
• ❌ Potential exposure of local files
• ❌ Malicious scripts could affect your computer
• ❌ Authentication tokens exposed locally

Operational Limitations:

• ❌ Requires your computer to remain online
• ❌ Can’t run when laptop sleeps or shuts down
• ❌ Performance impact on your local system
• ❌ Browser crashes affect your work
• ❌ Difficult to scale (limited to local resources)

​

The Cloud Browser Solution

Komo’s cloud browser architecture solves these problems through complete isolation and secure credential management: Enhanced Security:

• ✅ Complete isolation from your local system
• ✅ No access to your personal browser data
• ✅ Encrypted session management
• ✅ Sandboxed execution environment
• ✅ Secure credential storage separate from browser

Operational Benefits:

• ✅ Runs 24/7 without your computer online
• ✅ No performance impact on local machine
• ✅ Persistent sessions across executions
• ✅ 99.9% uptime reliability
• ✅ Scalable infrastructure

Modern AI agent architectures recognize that local browser automation introduces unnecessary security risks. Cloud-based browser infrastructure provides proper isolation where credentials are managed separately, sessions are sandboxed, and there’s zero risk to the user’s local system. This approach isn’t just more secure—it’s the foundation required for reliable, enterprise-grade AI operations.

​

How It Works

​

1. Agents Request Browser Access

When an agent needs to interact with websites: Task: “Research pricing for these 50 competitor products” Agent automatically:

1. Requests cloud browser instance
2. Receives isolated browser in <2 seconds
3. Navigates to competitor websites sequentially
4. Extracts pricing information
5. Compiles results
6. Closes browser (resources released)

No user intervention required—seamless and automatic.

​

2. Secure Session Management

For websites requiring authentication, Komo uses secure session management: First Time Login:

• Agent encounters login page
• Prompts you: “Log in to [website] to enable automation”
• You log in once through secure modal
• Session saved encrypted in cloud
• Future agents reuse authenticated session

Session Storage:

• Encrypted at rest and in transit
• Isolated per user (never shared)
• Time-limited and refreshable
• Revocable anytime via Settings

Session Persistence:

• Agents can access logged-in sites without re-authentication
• Sessions maintained across agent executions
• Automatic session refresh when needed

​

3. Manage Browser Sessions

View and control all saved sessions at Settings → Browser Sessions: Session Dashboard Shows:

• All websites you’re logged into
• Last used timestamp
• Session expiration date
• Which agents have access

Session Actions:

• Revoke - Delete session immediately
• Refresh - Re-authenticate if expired
• View agents - See which agents use this session
• Clear all - Wipe all sessions at once

​

4. Sequential Browser Execution

Agents use cloud browsers to complete tasks efficiently: Example: Product Research Task: “Research 100 products across 20 websites” Agent executes:

• Opens cloud browser instance
• Visits each website sequentially
• Extracts data systematically
• Maintains session across sites if logged in
• Compiles complete dataset

Reliability:

• Cloud infrastructure ensures uninterrupted execution
• Browser state maintained throughout task
• Automatic retry on transient failures

​

Real-World Examples

​

Example 1: Automated Competitive Pricing Monitoring

Scenario: E-commerce team tracks competitor pricing daily Workflow: Scheduled: Daily at 6 AM Agent automatically:

1. Opens cloud browser instance
2. Visits 15 competitor websites sequentially
3. Extracts current prices for tracked products
4. Compares against yesterday’s prices in database
5. Flags significant changes (>10%)
6. Generates price change report
7. Posts to #pricing-intel Slack channel

Session Management:

• Some competitor sites require login (research subscriptions)
• Sessions saved during first run
• Subsequent runs use saved sessions
• No re-authentication needed

Result:

• Complete pricing intelligence every morning
• Zero manual work
• Historical price tracking
• Immediate notification of competitor changes

​

Example 2: Lead Research & Enrichment

Scenario: Sales team needs to enrich lead data with publicly available information Workflow: Task: “For these 200 leads, research company websites and extract: company size, headquarters location, funding status, product offerings” Agent automatically:

1. Opens cloud browser
2. For each lead:
   • Searches company name
   • Visits company website
   • Navigates to About/Team pages
   • Extracts relevant information
   • Saves to structured dataset
3. Compiles enriched dataset
4. Updates CRM (Salesforce)

Session Security:

• No credentials needed for public company websites
• Data extracted securely in cloud
• Results delivered to CRM via API

Result:

• 200 leads enriched systematically
• Consistent data format
• CRM automatically updated
• No manual research required

​

Example 3: Automated Compliance Document Collection

Scenario: Legal team needs to collect regulatory filings from 100 companies Workflow: Scheduled: Monthly on 1st day Agent automatically:

1. Opens cloud browser
2. Navigates to SEC EDGAR system
3. For each company:
   • Searches company name
   • Finds latest 10-K filing
   • Downloads document
   • Renames file systematically
4. Uploads all files to Notion database
5. Updates compliance tracking spreadsheet

Benefits of Cloud Browser:

• Runs overnight (no computer needs to be on)
• Reliable execution (cloud infrastructure)
• Systematic organization
• Auditable (complete log of downloads)

Result:

• 100 filings collected automatically
• Organized systematically
• Compliance team reviews, not collects

​

Example 4: Multi-Site Order Monitoring

Scenario: Operations team monitors order status across 5 supplier portals Workflow: Scheduled: Every 4 hours Agent automatically:

1. Opens cloud browser
2. Logs into first supplier portal (saved session)
3. Checks order status for all pending orders
4. Extracts tracking numbers
5. Repeats for remaining 4 supplier portals
6. Compares against expected delivery dates
7. Flags delays (>2 days late)
8. Updates internal operations dashboard
9. Sends Slack alert if critical delays

Session Management:

• 5 different supplier login sessions saved
• Each portal has different auth method (handled automatically)
• Sessions refreshed as needed
• Operations team never logs in manually

Result:

• Real-time order visibility
• Proactive delay detection
• No manual portal checking
• Unified dashboard across suppliers

​

Example 5: Job Market Intelligence Gathering

Scenario: Recruiting team monitors job boards for candidate pipeline insights Workflow: Scheduled: Daily at 8 AM Agent automatically:

1. Opens cloud browser
2. Searches job boards (Indeed, LinkedIn, Glassdoor)
3. For each board:
   • Filters: “machine learning engineer” + “San Francisco” + “posted last 24 hours”
   • Extracts: company, title, salary range, requirements
4. Identifies competitors’ job postings
5. Tracks hiring trends over time in /workspace/hiring_trends.db
6. Generates weekly hiring intelligence report

Cloud Browser Advantages:

• Job sites often block automated scrapers
• Real browser = appears as legitimate user
• Sequential execution = respectful of site resources
• Persistent sessions = no repeated logins

Result:

• Comprehensive talent market intelligence
• Competitor hiring insights
• Proactive candidate sourcing
• Data-driven recruiting strategy

​

Security Architecture

​

Isolation Layers

User Isolation:

• Each user has separate browser pool
• No cross-user access or data sharing
• Independent encrypted session storage

Agent Isolation:

• Each agent execution gets fresh browser
• No cross-contamination between tasks
• Automatic cleanup after completion

Network Isolation:

• Browsers run in sandboxed containers
• Limited network access (web only)
• No access to Komo internal systems

​

Session Security

Encryption:

• All session data encrypted at rest (AES-256)
• Encrypted in transit (TLS 1.3)
• Encryption keys rotated regularly

Access Control:

• Sessions accessible only to your agents
• Time-limited access tokens
• Automatic expiration and cleanup

Audit Trail:

• Complete log of session usage
• Which agent accessed which site, when
• Export logs for compliance

​

Credential Protection

Zero Knowledge Architecture:

• Komo never sees your passwords
• You log in through browser directly
• Session cookies stored encrypted
• No credential transmission to Komo servers

Comparison to Local Browser:

​

Compliance

Certifications:

• SOC 2 Type II certified
• GDPR compliant
• CCPA compliant
• HIPAA-ready for enterprise

Data Retention:

• Session data: Until revoked or expired
• Browser history: Not stored
• Downloaded files: Saved to /workspace/temp/ with 7-day retention (Enterprise: custom retention via [email protected])

​

Managing Browser Sessions

​

View Active Sessions

Navigate to Settings → Browser Sessions to see: Session List:

• Website/service name
• Last accessed timestamp
• Expiration date
• Status (active/expired)

Session Details:

• Which agents use this session
• Access history
• Session cookie metadata (technical)

​

Revoke Sessions

Individual Revocation:

• Click session in list
• Click “Revoke Session”
• Confirm
• Session deleted immediately

Bulk Revocation:

• Click “Clear All Sessions”
• Confirm deletion
• All sessions wiped

When to Revoke:

• Changed password on website
• No longer need agent access to site
• Security concern or suspicious activity
• Offboarding employee (enterprise)

​

Refresh Expired Sessions

When session expires:

• Agent notifies: “Session for [website] expired”
• Click “Refresh Session”
• Log in again through secure modal
• New session saved
• Agent resumes automatically

​

Monitor Session Usage

Session Activity Log:

• View which agents accessed sessions
• Timestamp of each access
• Actions performed (page views, form submissions)
• Download activity log (CSV)

​

Best Practices

​

Session Management

Log in Early:

• ✅ Log into frequently-used sites before creating agents
• ✅ Test login works before scheduling automation
• ❌ Don’t wait for agent to fail mid-execution

Regular Review:

• Review active sessions monthly
• Revoke unused sessions
• Refresh expiring sessions proactively

Security Hygiene:

• Use strong, unique passwords for automated sites
• Enable 2FA where possible (handle during initial login)
• Revoke sessions when changing passwords

​

Browser Automation Design

Be Specific: ✅ Good: “Navigate to example.com/products, filter by ‘Electronics’, sort by price descending, extract top 20 product names and prices” ❌ Vague: “Get products from that website” Handle Variations:

• Account for different page layouts
• Handle loading states (wait for elements)
• Plan for error states (page not found, timeout)

Respect Rate Limits:

• Don’t overwhelm websites with requests
• Add delays between actions if needed
• Be respectful of website resources

​

Debugging Browser Automation

View Browser Screenshots:

• Agent captures screenshots during execution
• Stored in /workspace/temp/ (7-day retention)
• Review in activity monitor
• Identify where automation failed

Check Session Status:

• Verify session still active
• Check if login required
• Refresh expired sessions

Review Error Logs:

• Browser errors (timeout, element not found)
• Network errors (connection failed)
• Authentication errors (session expired)

​

Common Questions

Q: Does Komo access my local browser? A: No. Komo uses exclusively cloud-based browsers. Zero access to your local browser, history, or data. Q: How do I log into websites for automation? A: When agent needs authentication, you log in once through secure modal. Session saved encrypted for future use. Q: Where are my login sessions stored? A: Encrypted in Komo’s secure cloud infrastructure. Accessible only to your agents. Revocable anytime. Q: Can agents see my passwords? A: No. You log in directly through browser. Only session cookies stored (encrypted). No password exposure. Q: What happens if I change my password on a website? A: Saved session becomes invalid. Agent notifies you. Log in again to save new session. Q: Can I use cloud browser for banking/financial sites? A: Technically possible, but not recommended for highly sensitive financial operations. Use for research, public data only. Q: Can I run multiple browser instances simultaneously? A: Currently, each agent uses one browser instance at a time. Multiple agents can run in parallel, each with their own browser. Q: Do websites detect cloud browser as automated? A: Cloud browsers appear as real Chrome browsers. Most websites cannot distinguish. However, some security-sensitive sites may prompt additional verification. Q: What if a website blocks cloud browser IPs? A: Rare, but possible. Contact support for options. Q: Can I watch agents browse in real-time? A: Yes. Activity monitor shows screenshots as agent navigates. Screenshots saved to /workspace/temp/ with 7-day retention. Q: How long do browser sessions last? A: Sessions persist until: you revoke them, they expire (site-dependent, typically 30-90 days), or you change password on site. Q: Can multiple agents share the same browser session? A: Yes. All your agents can use sessions you’ve saved. Sessions reused efficiently across agents. Q: What happens to browser screenshots and downloads? A: Saved to /workspace/temp/ with 7-day automatic retention. Enterprise users can request custom retention by contacting [email protected].

---

Together, Cloud Compute Environment and Cloud Browser Automation provide the secure, scalable foundation for autonomous AI operations—running reliably 24/7 without accessing your local system, handling complex workflows, and maintaining enterprise-grade security throughout.