Custom-drawn controls
Buttons and grids painted straight to the screen look like controls to a person and like a blank rectangle to an accessibility API.
Automate the desktop apps your tools can't see
Custom-drawn grids, terminals and remote sessions break coordinate scripts. The Qventis Engine reads each control by sight, like a person, and records every click as a plain-English step that compiles once and runs the same way every time.
- Citrix, remote desktop and terminals
- No app source needed
- Heals approved by people
Open the 'Ledger' app
Choose 'Journal' from the 'Entry type' list
Set cell 'Debit' to '{{entry.amount}}' on row 1
Click the 'Post' button
Check that the 'Balance' label shows '0.00'
Where desktop automation breaks down
WPF and WinForms panes
Third-party and older components expose little to the tree, so scripts fall back to coordinates and break.
Aging drivers
Desktop drivers that see little active development leave teams with scripts that nobody can fix. Those suites need a path forward.
How the Qventis Engine sees every control
Vision recognition
The engine looks at the screen the way a tester does. It finds the 'Post' button by what it shows and where it sits.
- Works on custom-drawn grids and panes
- Reads terminal and packaged-app screens too
- Zero AI tokens at run time
Recorder and repository
Plain-English steps appear as you click. Each control gets one stored description that every test shares.
- Read a recording back as sentences
- Fix a changed control in one place
- Heals arrive with evidence for review
Remote sessions
When an app arrives through a remote session or a terminal emulator, the test machine sees only pixels. Vision recognition works there too.
- Citrix, remote desktop and terminal sessions
- The same steps as a local app
- Screenshots and values on every check
A typical first rollout
Start with one legacy app, then extend the same App Model to every channel.
- Record a flow testers run by hand
- Add checks and shared components
- Run headless from the CLI in your pipeline
- Review evidence in Quality View
Questions
Do we need source code or hooks in the app?
No. Vision recognition works from what is on screen, so the app stays unchanged.
Is AI used to find controls during a run?
No. Recognition is compiled into the test ahead of time and runs deterministically.
Can we keep our existing web framework?
Yes. Start with desktop alone and call the engine from your current framework through the CLI.
See your hardest desktop app automated live
We record a flow with you and run it from the CLI.