QBotRPA · AI-powered automation

Reliable execution.
A more intelligent way to build.

Describe the task. Review the workflow. Put it to work. QBotRPA combines visual automation with AI-assisted generation, diagnosis and controlled recovery across browser, desktop and data workflows.

Conversation + visual designBrowser + desktop executionReviewable. Traceable. Supervised.
QBot AI workflow design: conversation, canvas and node properties
QBot AI workflow design: conversation, canvas and node properties

Built for the work between systems

Connect the work
between your systems.

Keep your business systems in place. Bring their repetitive steps into a workflow that business teams can understand, specialists can maintain and IT can oversee.

01

Browser & desktop automation

Connect web interactions, desktop applications and file operations without repetitive manual handoffs.

  • Browser and desktop element picking
  • Ordered locator strategies
  • Stop when the target cannot be identified reliably
02

Visual workflow design

Make steps, branches and loops visible. Build workflows that can be inspected, tested and reused.

  • Node parameters and variables
  • Subflows and reusable steps
  • Separate testing, saving and publishing
03

Run, schedule & observe

Launch and schedule tasks, then follow outcomes through node traces, logs and optional recordings.

  • Task scheduling and run status
  • Node-level logs and run replay
  • Exception notifications, including email and WeCom

From instructions to inspectable workflows

Not just a chat window.
A different way to build workflows.

From a natural-language brief to a workflow draft, targeted edits and failure analysis, AI becomes part of the design process. The important part is not simply generating steps: it is understanding and validating them before execution.

Explain this export flow. Where might it fail?
Illustrative interaction with preset examples. No model calls or real workflow execution.

Understand before you change

  1. Identify page opening, waiting, export and file output.
  2. Explain what each selected step expects.
  3. Separate observed evidence from possible causes.

Explanation does not change the workflow.

01

Describe

Define the goal, inputs, outputs and permitted actions.

02

Review

Inspect changes and complete locators and dependencies. A draft is not a production release.

03

Test

Validate in a test environment and explicitly confirm actions with business side effects.

04

Publish

Use the established save and release process, with traceable change boundaries.

Adapt with control

From stopping at an error
to controlled recovery.

Page changes, loading delays or broken locators can interrupt stable workflows. AI uses execution context and available evidence to help diagnose the issue, propose a recovery path and attempt recovery within an authorized policy.

The goal is less repetitive troubleshooting, not fewer business controls. When evidence is insufficient, the target is ambiguous or the outcome is uncertain, stop and hand over to a person.

An operating workspace, not a collection of scripts

Clarity does not end
when a workflow is built.

See frequently used workflows, recent runs, exceptions and runtime status in one workspace. Move beyond building automation to understanding what it does each day and what needs attention.

QBot workspace overview — illustrative interface
QBot workspace overview — illustrative interface

Start with one useful workflow

Start with work that happens every day.

These illustrative scenarios are adapted to your systems, fields and permissions. Establish one reliable workflow before expanding into cross-team coordination.

COMMERCE

Commerce operations

Collect store and ERP reports, organize order, stock and fulfillment data, and route exceptions for review.

FINANCE

Reconciliation & documents

Collect statements and transaction records, match by rules and surface discrepancies. AI helps interpret unstructured material; people confirm important amounts and accounting decisions.

OPERATIONS

Recurring operations

Schedule file preparation, report updates and result distribution. Automate routine steps while retaining human ownership of exceptions and decisions.

QBotRPA + NovaPilot

AI understands and orchestrates.
RPA carries out the steps.

QBotRPA focuses on executable, manageable automation. NovaPilot brings together AI applications, digital workers, data and business collaboration. Together they connect analysis and recommendations to permissioned, traceable execution.

UNDERSTAND

AI applications: understand

Interpret material, extract information, generate content and prepare inputs for the next step.

EXECUTE

QBotRPA: execute

Operate browsers, desktop applications and files through defined steps, returning status and available evidence.

COORDINATE

Digital workers: coordinate

Coordinate tasks around role-specific goals, follow exceptions and hand over decisions that require people.

A few things worth knowing

Before your first workflow.

Can an AI-generated workflow run in production immediately?

Generated workflows are proposals for review. Check nodes, locators, inputs, credentials and scope, then test before saving and publishing. Actions with business side effects still require explicit authorization.

Does self-healing eliminate workflow maintenance?

No. AI can help with some environmental changes and failures, but cannot guarantee an understanding of every business rule. Major system changes, complex permissions, duplicate-action risks and uncertain outcomes still need human maintenance.

Must existing workflows be rewritten for AI?

No. Keep rules and deterministic nodes for stable operations. Add AI where understanding, generation or diagnosis is useful. Model availability is distinct from ordinary visual editing and established execution capabilities.

How do I start?

Visit the download page and check the installer’s supported system. Contact the team to confirm trial activation, runtime requirements and a first scenario. Start with redacted samples and test accounts.

Start with one task worth automating.

Bring a repetitive task. Together, we will define the inputs, steps, exceptions and acceptance criteria.