THE ASSESSMENT LAYER FOR VIRTUAL LABS

The expert grader that never blinks.

Send or stream your lab’s video to one simple API. Powerful vision models can handle any OS, GUI or CLI and return results beyond just completion. No access to your environment needed.

PIPELINE

Video in. Verdict out.

/ 01

Send the signal

Point your lab’s video output at the API: upload a finished recording or stream the session live. Pixels are all we need.

/ 02

The models watch

Powerful vision models follow the session end to end, reading the screen the way an expert proctor would, step by step, in context.

/ 03

Get the verdict

A structured result comes back: completion, effort, points of struggle and integrity signals, ready for your platform.

COVERAGE

Any screen. Every stack.

Because Vidience reads video, not internals, it covers every common lab environment. No dependency on how it’s built.

Native GUI

Desktop applications, rendered anywhere.

Web GUI

Browser consoles, dashboards and web IDEs.

Any OS

Windows, Linux, macOS. Wherever the lab runs.

Domain CLIs

Domain-specific command lines and terminal-first workflows.

Zero
access
required

Vidience never enters the client environment. Nothing to install, nothing to integrate inside the lab. Video out → results back. That is the entire integration surface.

NATIVE GUI WEB CONSOLE WINDOWS LINUX MACOS NETWORK CLI DATABASE SHELL TERMINAL
DEPTH

Beyond the checkbox.

Completion tells you the box got ticked. Vidience tells you what it cost: the effort spent, the moments of struggle, and whether the work was honest.

EFFORT

How hard the learner actually worked, not just whether they finished.

STRUGGLE

The exact moments where things got hard, surfaced so you know where learners fight the material.

INTEGRITY

Every session is watched for attempts to cheat, flagged as the models follow along.

SESSION REPORT — asmt_91f3 WATCHING
COMPLETED
true
EFFORT
0.00
STRUGGLE
0 pts
INTEGRITY
clear
00:00 EFFORT OVER SESSION 47:12
12:40 — repeated failures on step 3 31:05 — long hesitation before rollback 0 integrity flags
WHO IT'S FOR

Now grading: you.

Four checks. If they land, Vidience was built for your platform

FIT ASSESSMENT
[]

You run a practice, cert-prep, or hands-on lab platform.

[]

You maintain in-VM grading scripts, and your engineering team feels it.

[]

You ship new labs faster than your grader pipeline can keep up.

[]

A security review has pushed back on what runs inside your virtual environments.

VERDICT — FIT CONFIRMED GET EARLY ACCESS →
FAQ

Asked, answered.

Can you grade my lab format?

If a human reviewer could grade it from a screen recording, yes.

What model is under the hood?

Frontier vision models behind a swappable assessor interface. When better or cheaper options ship, they slot in without changing the API contract.

Where does the video live?

Encrypted blob storage in your region. The default container purges within 24 hours; JSON verdicts are retained 30 days.

What about EU data residency?

Supported. Storage and processing both run in your region, so EU data stays in the EU.

How does pricing work?

Per-assessment, billed monthly. We size it against your volume during the intro call. Pricing is being calibrated with partners and isn't published yet.

EARLY ACCESS

Your labs already produce the video. Let it speak.