Home / Platform
Detect. Review. Redact. Prove.
Most redaction software stops at the third step. The fourth is the one that decides whether a release survives an appeal, and it is where Redactor is built differently.
At a Glance
315
Distinct file extensions across five content types
12
Compliance frameworks, each with its own class set
6
Deployment postures, shared SaaS through air-gapped
36
Audited event types, ingestion to disposition
The Four Stages
Detect
Stage 1
The file is analysed for the object and PII classes you selected. Detections are stored as timed data indexed by position over time, so the studio shows where each object appears across the timeline rather than frame by frame. Runs at upload or on demand, scoped by duration and by class.
Learn moreReview
Stage 2
Nothing releases on the detector's say-so. An operator merges duplicate tracks, adjusts boxes, deletes false positives and draws what was missed. Redactions organise into named layers, so one asset serves several disclosure tiers without separate copies that drift.
Learn moreRedact
Stage 3
The redaction is burned out of a new rendition rather than layered over the original, and what happens to the source is a configured policy with three options. For volume the same pipeline runs with no operator opening the file.
Learn moreProve
Stage 4
Every redaction carries the exemption it was made under, printed onto the mask and into the transcript. Coverage reporting states what share of a framework's detection set was caught, and the custody trail records every action with user, email, IP and timestamp.
Learn moreAutomated detection proposes; a person disposes. That step is what makes the output defensible, and it is the half of the job the category mostly ignores.
Across Every Format
Video Redaction
44 formats
Faces, persons, plates, vehicles, weapons, screens, tattoos, on-screen text and custom objects, tracked across frames and through 360 degree footage.
Learn moreAudio Redaction
11 formats
Spoken identifiers found by reading the transcript rather than listening, with speaker separation so the right voice is cut.
Learn moreDocument Redaction
63 formats
Text, objects embedded inside PDFs, tables by row and column, non-Latin scripts, and scans that arrive rotated or badly photographed.
Learn moreImage Redaction
187 formats
Identity documents, faces, plates and screens across large batches, including 360 degree panoramas.
Learn moreDICOM Redaction
Medical imaging
Pixels and header tags de-identified as separate operations, every frame of a multi-frame study, output stays DICOM.
Learn moreWhat Runs Underneath
Detection Classes
Every object and PII class the detector recognises, with confidence thresholds selectable per job.
Learn moreOCR and Text
Text lifted off images, screens, scans and embedded objects, which is what makes a photographed page redactable at all.
Learn moreFile Formats
The full list of 315 extensions, what each is rendered by, and what comes back out.
Learn moreAutomation and Scale
Volume
Queues, reusable templates, per-format permissions and headless processing for work nobody opens by hand.
Learn moreReview and Approval
Governance
The human checkpoint, redaction layers and disclosure tiers.
Learn moreDefensibility
Proof
Exemption codes, coverage reporting, custody and the redaction record.
Learn moreREST API
Integration
The same endpoints and OpenAPI spec in every deployment posture.
Learn moreWhere It Runs
Shared SaaS
Multi-tenant, isolated by tenant identity
Dedicated cloud
Isolated tenancy, operated by VIDIZMO
Private cloud
Your own cloud subscription
On-premises
Your data centre, your hardware
Hybrid
Cloud application reaching on-premises directories and storage
Air-gapped
Disconnected from the public internet entirely
The same engine runs in all six. AI inference is self-hostable, so detection need not call out of your network, and in an air-gapped deployment the whole stack is local: language models through vLLM or Ollama, plus local models for transcription, vision, OCR and PII detection.
FAQ
The Redactor Platform questions, answered
What are the four stages of redaction?
Detect, review, redact and prove. The file is analysed for the object and PII classes selected, an operator confirms or corrects each detection, the redaction is burned into a new rendition, and the statutory exemption is recorded against it. Most redaction software stops at the third stage.
Does redaction happen automatically or does someone review it?
Both. Automated detection proposes and a person disposes. Detection handles the volume, and review is where duplicate tracks are merged, false positives deleted and anything the detector missed is drawn by hand. For a consequential release the review step is what makes the output defensible.
What file formats does the platform handle?
315 distinct extensions across five content types: 187 image, 63 document, 44 video, 13 miscellaneous and 11 audio. Three rendering paths handle them, so a format is ingested rather than converted first.
Is the original file kept after redaction?
That is a configured policy with three options rather than a side effect. The default creates a redacted copy and retains the original, which is what public records and evidence work require, because an appeal against a withholding is decided against the unredacted record.
Can the same engine run on-premises?
Yes. The same engine runs in shared SaaS, dedicated cloud, private cloud, on-premises, hybrid and air-gapped deployments. AI inference is self-hostable, so detection need not call out of your network.
See It on Your Own Files
Send us something difficult and we will run all four stages on it.