CUSTOMSCREENCOMPUTER-USE
Software Task Screenflows
Screen workflow recordings with task objectives, OS/app metadata, cursor activity, step boundaries, completion status, and action logs for computer-use model evaluation.
CUSTOMSCREENNARRATED WORKFLOWS
Guided Software Walkthrough Sessions
Narrated software workflow collection format for procedural AI, visual-language models, instruction following, and software assistant evaluation.
Guided Software Walkthrough Sessions are built for teams that need screen activity paired with human explanation. The format captures contributors completing software tasks while narrating what they are doing, why they are doing it, and how the workflow progresses.
The result is a workflow dataset that connects screen state, user intent, and spoken explanation.
Structured collection
Guided Software Walkthrough Sessions use defined software tasks and narration instructions.
Metadata depth
The format supports alignment between video timestamps, narration utterances, step markers, cursor events, and task phases.
Delivery-ready package
Includes screen clarity, narration clarity, transcript alignment reviews, plus dataset manifest, CSV catalog, consent status…
Request sample access
Share your target software workflow, narration requirements, transcript depth, and evaluation goals. We’ll reply with a technical scope for the collection.
Key highlights
- Screen recordings paired with aligned narration transcripts.
- Step markers connected to cursor flow, app context, and workflow state.
- Metadata for software category, OS, task type, narration language, and completion status.
Dataset contents
Screen capture and logs
Screen recording files · Audio narration track · Narration transcript files · Step markers
Task and workflow records
Task objective metadata · Software category metadata · Operating system metadata · Cursor flow metadata where scoped
App and session metadata
App context metadata · Completion status
Technical specifications
Use cases
- Procedural AI evaluation
- Instruction-following model testing
- Visual-language model training
- Software assistant evaluation
- Screen-to-language grounding
- Workflow explanation modeling
- Task decomposition research
Dataset workflow
1Collection
Guided Software Walkthrough Sessions use defined software tasks and narration instructions. Contributors complete a workflow while describing the steps, decisions, or interface changes in real time or through guided narration.
2Review
Screen clarity review · Narration clarity review · Transcript alignment review · Step marker review · Task match review · Consent status review · Delivery approval
3Delivery
MP4 screen recordings · Narration audio where separated · Narration transcript JSON · Step marker JSON · Task metadata CSV · Dataset manifest JSON · CSV file catalog · QA summary · Consent status metadata · Delivery notes
Quality assurance
Related dataset formats
CUSTOMSCREENWEB WORKFLOWS
Browser Navigation and Web Task Flows
Browser task recordings with URL context, click and scroll events, form interaction metadata, page categories, task outcomes, and action logs for web-agent evaluation.
CUSTOMSCREENMOBILE WORKFLOWS
Mobile App Task Screenflows
Mobile screen recordings with tap events, swipe events, app categories, task objectives, step boundaries, completion status, and device metadata for mobile-agent evaluation.
CUSTOMSCREENERROR RECOVERY
Error Recovery and Correction Screenflows
Screen recordings focused on failed attempts, undo actions, retries, correction paths, error markers, recovery steps, and completion outcomes for agent robustness testing.
CUSTOMMULTIMODALPAIRED DATA
Multimodal Paired Capture Sets
Paired multimodal datasets linking video, audio, screen, text, or image-derived records through shared file IDs, cross-modal manifests, metadata schemas, and QA fields.
CUSTOMANY MODALITYBUYER-SCOPED
Buyer-Scoped Custom Collection
Fully scoped custom data collection with buyer-defined modality, schema, metadata fields, annotation layers, QA criteria, manifest structure, delivery package, and collection rules.