AuViScan
Open beta · bootstrapped from Enugu, Nigeria

Moderation that
understands context.

AuViScan analyzes video, audio, and on-screen text together, flagging violations with explanations your team can actually use. A car crash with news commentary is not gratuitous violence, and your reviewers should not be asked to argue otherwise.

Beta users

947

Videos analyzed

5,600+

Avg processing

13.4s

analyzing · live
job_3f2c · 00:00:48
Audio
Visual frames
On-screen text
Verdict pending
Compatible with
  • AWS
  • Amazon IVS
  • Brightcove
  • Vimeo
  • Custom API
How it works

One job ID. Five steps. Real reasoning at the end.

Submit a video, get back a verdict your trust-and-safety team can act on in under 30 seconds.

  1. 01

    Submit video

    Upload via API or a pre-signed S3 URL. The job enters the queue with a returned jobId.

  2. 02

    Multi-modal analysis

    Frames, audio transcript, and on-screen text are analyzed together. Context is the unit, not the modality.

  3. 03

    Verdict + explanation

    A PASS / FLAGGED / BLOCK verdict comes back with the natural-language reasoning your team can quote.

  4. 04

    Human review

    Flagged items enter a prioritized queue. Reviewers see AI pre-notes, cited timestamps, and the original audio.

  5. 05

    Highlights (optional)

    Brand-safe compilations are generated automatically from passed content, turning moderation into revenue.

Why AuViScan

Built for the reviewers, not the demo deck.

Live numbers from the open beta. Updated weekly as the 947-user cohort runs more videos through.

87.2%

Detection at sports-context fidelity

Measured against Rekognition alone (78%) across our 947-user open beta. Numbers update as more verdicts land.

Cite the reason

Every flag includes its evidence

Natural-language reasoning, timestamps, modality breakdown. No black-box verdicts.

Highlights

Moderation becomes revenue

Brand-safe clips compiled from passed long-form video, scoped by your existing policy.

Saves 18%

Fewer false positives, less reviewer time

Context-aware verdicts mean reviewers stop arguing with the bot and start clearing the queue.

Videos in open betaLive

5,634+

Counter updates as beta users push jobs through the demo

Average processing

13.4s

Across all video lengths. Will drop sharply once AWS scale credits land.

False positive rate

8.1%

Down from 9.6% on Rekognition-only baseline

Beta user · West African sports streaming team

“Before AuViScan, our reviewers were spending 25 minutes per flagged video. Now they spend under nine. The explanation tells them exactly what to look for.”

Head of operations

A leading West African sports streaming platform (name withheld during pilot)

Open beta

Free until Q4 2026.

The open beta is free. 947 people are on it. Three platforms signed early enterprise contracts and pay us about $5,800 a month, which keeps the AWS bill paid while we wait on Activate Portfolio credits.

  • Free, unlimited, no card required during open beta
  • Beta users get founding rates locked at -50% when paid tiers launch
  • Enterprise contracts available today, on direct conversation
Try without an account

See your first verdict in
under ten seconds.

The demo runs entirely in your browser. Drop a clip, watch the verdict assemble, scrub the timeline. No signup, no data leaves your device.