CAPTURE
Work naturally in a real webpage
Clicks, typing, navigation, and waits become an editable sequence while you complete the task.
Browser Agent
Allow Everyone To Automate
Record browser work once. Turn it into reliable, scalable automation—without writing scripts.
Browser Agent is available in Free, Advanced, and Enterprise editions.
All plansFree, Advanced, and Enterprise
3 build methodsRecord, extract, or design visually
CSV · JSON · XLSXStructured export formats
Live execution recordsStatus, step logs, and results
Three ways to create automation quickly
Record a real task, generate a workflow through the extraction wizard, or build from a blank visual canvas. Browser Agent turns team know-how into maintainable workflows.
RECORD & REPLAY
Record browser actions
Complete a task once to capture clicks, inputs, navigation, and waits as workflow steps that remain editable and reusable.
1
Start recordingComplete the task in a real webpage
2
Review stepsAdjust selectors, waits, and input values
3
Replay reliablyRun on demand and reuse the workflow
Visual workflow editor
Drag browser, interaction, condition, loop, and data blocks onto a canvas; add variables, sub-workflows, and parallel branches for complex processes.
Structured extraction wizard
Select fields, tables, repeated blocks, and detail pages visually; configure pagination and limits, then output CSV, XLSX, sheets, or email results.
Four Core Pillars
Everything required to build, run, and control browser automation
Start with the four product pillars, then expand a panel whenever you need the detailed capabilities.
01Build visuallyTurn browser work and business rules into maintainable flows.
Record and replayDrag-and-drop editorExtraction wizardConditions and loops
02Run and scaleExecute one workflow or coordinate recurring work at scale.
Manual and batch runsSchedulesParallel tab groupsPage-change monitoring
03Process data with AIMove structured web data directly into repeatable AI processing.
Local sheetsField binding and write-backLLM batch processingCSV, JSON, and XLSX
04Monitor and control securelyObserve every run and keep sensitive inputs within clear boundaries.
Live status and logsBreakpoints and step modeCredentials and MFAIsolated authorized sessions
Interactive Product Tour
From a recorded task to a monitored run
Select a stage to see how Browser Agent turns real browser work into reusable automation.
REFINE
Add the logic your process needs
Review steps, bind variables, and add conditions, loops, data operations, or reusable sub-workflows.
RUN
Replay, observe, and improve
Execute in the assigned browser session, follow progress, and use run records to refine the workflow.
Parallel Tab Groups
Multiple tab groups. Multiple workflows. All running in parallel.
Isolate tasks, data, and sessions. Run unlimited automation workflows at the same time without interference, with every workflow executing in its own browser tab group.
Isolated tasks
Each workflow keeps its own tabs, session, and execution state.
True parallel execution
Start independent workflows together instead of waiting in a queue.
Organized by group
Group tabs by project, client, account, or automation task.
Live visibility
Follow progress, current step, runtime, and results for every workflow.
Private Automation
One tab group. Private automation. Fully controlled.
Run authorized automation inside an isolated browser profile with controlled proxy, fingerprint, and session settings.
Private browser environment
Keep automation activity inside its assigned browser profile and session boundary.
Proxy routing
Assign a proxy and region to each session, with connection status visible while it runs.
Custom fingerprints
Reuse controlled browser profiles for a predictable, consistent session identity.
Isolated sessions
Keep cookies, local storage, and task state contained within the assigned tab group.
LOCAL SHEET × LLM
Scrape it once. Let an LLM process the local sheet.
Once web data lands in a Browser Agent local sheet, there is no export, copy, or tool switching. Select columns, describe the task in one prompt, then classify, summarize, translate, clean, or extract fields across every row and write the results back to new columns.
01
Select your columnsReference titles, descriptions, reviews, or any scraped field
02
Describe the resultUse plain language for the decision, transformation, and output format
03
Run in bulk and write backProcess local rows and save structured results into new columns
#ProductDescriptionAI summaryCategory
Travel tumblerDouble-wall vacuum...Compact, leakproof commute cupOutdoor
Desk organizerModular tiered design...Flexible modular desk storageHome
Wireless fill lightThree color modes...Portable light for live videoCreator
4Packing cubesWater-resistant set...Waiting to process—
✦
LLM bulk processApply to 128 selected rows
Prompt
Read {{Description}}, write a summary under 20 words, and return one product category.
SummaryWrite to “AI summary”
CategoryWrite to “Category”
Processing row 96 of 12875%
Security and execution controls
Run only on pages and data you are authorized to use. Keep credentials separated, retain execution records, and respect site terms and applicable law.
DUO KAI BROWSER
Courier
Courier
Universal
Built for frequent, rule-based browser work
E-commerce back-office tasks, authorized data preparation, page checks, and change monitoring
Record product maintenance, order lookup, and form-entry steps; drive batch input from sheets and keep a record of each run.
E-commerce Operations
Standardize back-office workflows
Select fields, lists, tables, and detail-page information from pages your organization may access, then filter and export to files or sheets.
Business Data Preparation
Structured extraction and export
Check critical pages, elements, and interaction paths on a schedule, then record and notify when content changes or a flow fails.
Page Inspection
Change monitoring and web testing
Browser Agent vs. RPA
Browser-native automation or traditional desktop RPA?
Both automate repetitive work, but they use different operating models. Traditional RPA is built around desktops and software robots. Browser Agent brings browser actions, web data, isolated sessions, and parallel workflows into one connected workspace.
Desktop-centered
Traditional RPA
Best suited to native desktop software, legacy systems, and processes spanning several local applications.
Browser-native
DUOKAI Browser Agent
Best suited to web operations, structured web data, multi-account sessions, monitoring, and parallel browser tasks.
Capability
Traditional RPA
Browser Agent
Workflow creation
Vendor studios, UI selectors, scripts, and platform-specific components
✓Record real browser actions, then refine them visually with conditions, loops, and data steps
Runtime environment
Managed desktops, virtual machines, runners, and application state
✓Run directly in assigned browser profiles and tab groups with session context available
Parallel scale
Capacity commonly planned around robot licenses, runners, and queues
✓Run workflows in separate tab groups so tasks, data, and sessions remain isolated
Web data
May rely on UI scraping, OCR, scripts, or separate data components
✓Built-in structured extraction, local sheets, field write-back, and LLM batch processing
Monitoring and debugging
Orchestrator consoles, robot logs, screenshots, and desktop-session diagnostics
✓Live status, step logs, breakpoints, and step-by-step results in the workflow workspace
They can work together: use Browser Agent for browser execution and web data while traditional RPA continues to handle desktop software and legacy systems.
Bring Browser Agent to your team
See how Browser Agent can turn repeated browser work into reliable, scalable automation for your organization.
Frequently asked questions
What browser tasks can Browser Agent automate?
Browser Agent can orchestrate navigation, clicks, inputs, waits, tab operations, conditions, loops, parallel branches, data processing, exports, and notifications, with schedules and page monitoring. Use it only for authorized workflows that follow target-site rules.
How do I create a Browser Agent workflow?
Record browser actions, generate a flow through the extraction wizard, or drag browser, interaction, condition, and data blocks onto the visual editor.
How does Browser Agent handle extraction and output?
Select fields, tables, repeated blocks, and detail-page content from authorized pages, then export CSV or XLSX, update a data sheet, or send a completion notification.
How can I observe and troubleshoot workflow runs?
The execution center shows live state, step records, and results, while the visual editor supports breakpoints and step-by-step debugging.
