Begin with the material your business already owns
Martius works with local documents, product photos, footage, recordings, portraits, source audio and brand material. Original files and editable projects stay on your device; beginning a task does not move your source library into a cloud workspace.
Keep the project beside the actual business materials that support it. Select the sources needed for the task and retain their provenance, version and authorized use as the Agent creates the video.
Keep sources identifiable when files change
Source references preserve the identity, location, version and dependency of the material used in a project. A missing or replaced resource remains visible, and a moved file can be reconnected instead of silently becoming a different source.
Review a replacement before it enters an approved scene. If the product image, source procedure or recording changes, check the dependent facts, narration and review decisions rather than treating the old approval as current.
- Original file and source provenance
- Version, location and dependency
- Missing or replaced material requiring attention
- Authorization and scope for the selected use
Own the editable project after the first film
Keep the brief, scenes, approved references, presenter and voice choices, project revisions, exports and publication history in the local workflow. Reopen the project to update an offer or create another commercial task from the relevant approved material.
A finished video and an editable backup are different assets. A backup identifies the project structure, included materials, source references and external dependencies needed for restoration. Inspect that scope before moving the project to another device.
Choose the boundary of every delivery
Finished videos and course packages remain local by default. Local export writes the selected files to your device; publishing is a separate action for a named account and destination.
Approve the exact delivery copies and necessary metadata before sending. Publication does not include unrelated folders, original footage or the editable project. Learning events and destination analytics have their own selected data flow.
Local source files do not mean every AI operation is offline
Storage, generation and publication are separate questions. A workflow can keep the original file local while using model requests or derived data with an external service. Consult the workflow’s service and data scope before selecting that operation.
The local-project capability is not a claim of fully offline generation, private-model training, certification or a universal zero-network workflow. It gives your source materials, editable project and finished files a clear home and makes remote delivery an explicit choice.