Capture, Selector, and projects
Visual sources, timeline, markup, connections, descriptions, history of changes and saving the project.
Product status
What is confirmed in the current development state
In the current source, General and GUI Automation support the main workflows. Industrial and UAV are present as experimental modes, but do not yet form complete production workflows. The current file metadata does not establish the capability set of a specific published installer.
Program modes
Mode readiness
| Mode | Status | In source | The main process is ready | Current boundary |
|---|---|---|---|---|
| General | Implemented | Yes | Yes | Markup, general computer vision techniques and project saving. |
| GUI Automation | Implemented | Yes | Yes | Preparation of interface data and separate execution of the tested script. |
| Industrial | Experimentally | Yes | No | Subject data types, method execution, validation, and confirmed markup changes are available. There is no completed production process. |
| UAV | Experimentally | Yes | No | You can work with frames, run methods, and check the results. Mission control, GIS and autopilot are not included in the mode. |
Opportunities
Readiness of main parts of the product
Computer vision methods
Classical methods, methods with models, sequences and graphs of methods, preview and saving of results. Some methods require separate models and dependencies.
Automation Runtime
Loading a project, searching for visual and textual states, waiting for conditions and acting through the selected input method. Running the script is always done separately.
Screph Code
The embedded code environment receives the selected task context. The agent can change files in the working directory, so changes must be checked and stored in a version control system or a separate copy.
Agent
The built-in assistant is disabled by default. Suggested changes are shown before implementation, and high-risk actions require confirmation.
Features and limitations →Action CV
Computer vision techniques help determine the purpose of a recorded action. The proposed result is stored separately and requires human verification.
Speech and LLM connections
Local options and connections to external services with their own credentials are supported. Availability may vary by provider, network, and model selected.
Voice2Text
Dictation, window bindings, text review, and optional screenshot processing with a language model are implemented. The Voice2Text page shows current download availability and file details.
Description Voice2Text →Account and support
The account is used for profile, desktop program connection, support and available online operations. No login required to work locally.
Package purchase
Payment on the site is disabled. It is currently not possible to top up the balance displayed in your profile through the website.
Dependencies
What may be required in addition
Models and calculations
YOLO, SAM, OmniParser and other model methods require suitable model files. Tesseract OCR requires local installation of Tesseract. A separate graphics card is not required for basic functions.
Network Services
Cloud speech recognition and LLM require network and access to the selected service. Local connections are available with installed models.
Input for automation
Interface actions require a configured Arduino HID or FakerInput Virtual HID. Installing the FakerInput system driver is a separate action that requires Windows confirmation.
Limitations
Boundaries of the current development state
- Supported platform - Windows 10/11 x64. Builds for Linux and macOS are not released.
- The Alpha version cannot be used for unobserved automation without testing, checking the results and being able to roll back.
- Projects are stored locally. Audio, images, or context are provided to the external service only in user-selected network operations.
- The Voice2Text page shows current download availability and file details.
- The balance shown in the profile cannot currently be topped up through the website.