Alternatives & Comparisons

How to Run a Same-SOP Test for AI Training Video Generators

Test the SOP you actually need to teach. Golpo turns that procedure into a lesson you can score for coverage, order, exceptions, and update cost. The embedded video is a Golpo demonstration from the public NIST CSWP 29 framework. The SOP pack is the test you run on your own procedure.

Sudip KarSeptember 8, 202611 min read

An evidence-led Golpo guide illustrating how to run a same-sop test for ai training video generators

Test the work you actually need to teach. A single SOP can show how well Golpo turns source material into a lesson that is usable, faithful to the procedure, and practical to update. Golpo drafts the narration from the document, you approve the script, and you score the first video on coverage and maintainability.

Two different artifacts sit on this page. The video below is a Golpo demonstration made from the public NIST Cybersecurity Framework (CSF) 2.0, NIST CSWP 29 (February 26, 2024). Use the SOP pack after it to test a real procedure. Other products are not scored here.

The Golpo demonstration: NIST CSWP 29

The source is the official 32-page PDF, NIST CSWP 29. It is a framework for outcomes, and it says the Functions are addressed concurrently, with GOVERN informing the other five. It is not an operating procedure. It has no trigger, no ordered workplace actions, no warning placed before a risky step, no exception path, and no escalation owner. That is why it cannot stand in for the same-SOP test.

Documented configuration

Source: complete NIST CSWP 29 PDF, linked above.
Output: this Golpo explainer, 2 minutes 26 seconds, Canvas Editorial Expanded.
What to compare: six Function names, GOVERN informing the others, concurrent Functions rather than a one-way sequence, and outcomes rather than a mandated implementation. Check those against the PDF.

The SOP the test actually needs

Use a permission-cleared procedure with a trigger, five to nine ordered actions, one warning, one exception, a named escalation owner, a visible success condition, and a revision date. The procedure below is hypothetical. It shows the shape of a fair test. Swap in the SOP you need to teach, and keep that file’s checksum with the result.

Missing company laptop Revision date: 2026-10-01 Owner: Security Operations Trigger: An employee reports that a company laptop is missing. 1. The employee stops using accounts that were signed in on that laptop. 2. The employee records the time the loss was noticed and the last known location. 3. Warning: Do not wipe, track, or remotely lock the laptop yourself. Security Operations decides that after the report is accepted. 4. The employee notifies Security Operations in the same business hour. 5. Security Operations logs the report and contains the accounts. Exception: If the laptop is found in a locked company room and has not left the building, the employee tells Security Operations and waits for a release before using it. Escalation owner: Security Operations lead. Success: The report is logged, and Security Operations confirms the accounts are contained.

Create a two-minute orientation for employees from the attached approved SOP. Preserve every required action in order. State the warning before any wipe, track, or lock step. Keep the locked-room exception distinct from the normal path. Name Security Operations as the escalation owner. Do not add deadlines, consequences, or tools that are absent from the source. End on the success condition. Show the source title and revision date on the final frame. Insert the real asset-report screen where the employee files the notice.

Run the test in Golpo

Source. Freeze the approved SOP and its checksum. Upload it in Prompt Mode with the brief above. Golpo drafts the narration from the document. Read that draft against the trigger, the five actions, the warning, the exception, and the owner. Uploading the file does not certify the draft.
Settings. After the obligations are right, render the approved sentences in Script Mode and put drawing notes in Video Instructions. About 1,050 characters of narration fit in one selected minute. Turn on on-screen text from script only so visible terms stay inside the SOP (“Security Operations,” “same business hour”). With Custom Scenes, give the warning its own scene before any wipe picture, and give the locked-room exception its own scene. Insert the real asset-report form with Own Media. A drawn picture of a form is not that screen.
Script review. Approve the script with Edit script before creating video, then render. Save this first output before anyone repairs it.
Result. Score the first cut with the sheet below. An operational owner checks the procedure. A learning reviewer checks whether a new employee could follow it.
One correction. If the narration is in order and the picture shows the wipe before the warning, fix that frame with frame editing. The first edited version is free across sessions until finalized. Later frame editing charges the selected duration and does not change the audio. If the spoken steps are out of order, or the label says “IT help desk” while the script says “Security Operations,” revise the script and render again. Write down the selected minutes and the credits for that charged version.
One reuse. Change one source step — the notify window, the owner, or the exception — and run the test again on that version. The point is whether you can find the affected scene, replace the published file, and keep the previous approved cut for audit.

What each control changes

Control	What it answers in the test	Plan
Document to script	Prompt Mode drafts narration from the SOP so the team is not storyboarding the procedure by hand first. The draft still gets an obligation check.	All plans. Prompt and Script Mode
Script Mode	The approved sentences are what get spoken, including the warning and the owner’s name. Visual directions stay in Video Instructions.	About 1,050 characters per selected minute
On-screen text from script only	Labels draw from the SOP wording and may show a subset. This is not a caption track and not a check that the SOP itself is correct.	Canvas and Sketch. On-screen text
Custom Scenes	You place the warning before the risky action and keep the exception off the normal path. The map does not rewrite the narration or remove the engine limit.	Growth and higher, and Pay As You Go. Custom Scenes
Own Media	The asset-report screen in the video is your screen.	Business and higher, and Pay As You Go
Frame editing	One charged visual correction after the first score. Audio stays as rendered.	All paid plans

Score the explanation, then the update

Score whether the lesson preserved the procedure and whether you can maintain it. Leave visual taste out of the score. This sheet is for the Golpo run. It is not a leaderboard of other products.

Dimension	Pass condition	Automatic block
Coverage	Every mandatory action appears	A required step is missing
Order	Dependencies stay intact	The risky action appears before the warning
Exceptions	The exception and the escalation stay distinct	The normal path is shown as the only path
Terminology	Narration and labels use the SOP’s names	“Security Operations” becomes a different team
Screen evidence	The inserted screen matches the narrated step	A generated interface stands in for the real form
Updateability	Source version maps to the output and the destinations	No way to find the stale copy
Accessibility	Captions, transcript, contrast, and the player are checked	A required fact exists only in the picture

Seven stages and the result log

Freeze the approved source and checksum.
Record the plan, the settings, and the exact brief.
Save the untouched first output.
Have the operational owner check procedure fidelity.
Have the learning reviewer check clarity and how much the learner must hold at once.
Record the work and the credits to an acceptable version.
Change one source step and repeat.

Give the reviewer the whole package: source title and checksum, role, learning outcome, brief, plan and settings, generated script, first output, repairs, final output, reviewer names, approval date, and every destination. Fill the log from your run, including the reviewer’s start and finish.

Field	What to write down
Selected duration	The minutes you selected for the test. The pack above uses 2.
First-output credits	1 credit per selected minute for a standard video. A 2-minute selection is 2 credits.
Review clock	Reviewer start and finish, in your own notes.
Correction	Frame edit or new script render, what changed, and the credits charged for that version. A charged 2-minute version is another 2 credits.
Source update	The one step you changed, the new asset, and the destinations you replaced.

Plan and credits for the full pack

The full pack — real screen plus a visual correction — uses Business or higher, or Pay As You Go, because that is where Own Media is available. Frame editing: all paid plans. Custom Scenes is on Growth and higher and on Pay As You Go, so a document-and-script test without the real screen can start there. A standard 2-minute video is 2 credits. A later edited version is charged again on the selected duration. On Pay As You Go, edit-script-before-video bills 10% after the script and 90% after the video. Current amounts are on pricing. The Free plan is not a downloadable publishing workflow. If you later generate through the API, store the job id next to the source checksum in your own log. Scale includes API access, Business has a paid API add-on, and API Only is metered separately.

Review checklist

The source is the SOP you teach, with a trigger, ordered actions, a warning, an exception, and an escalation owner.
The NIST video is identified as the CSWP 29 demonstration.
First output saved before repair.
Warning before the risky step, in narration and in the picture.
Labels match SOP terms. The real screen is the inserted screen.
Credits and reviewer time written down for the first version and the correction.
One source change produces a replaced asset and a kept prior version.
Captions, transcript, and player checked before anyone calls the lesson ready.

Generate your first module from the SOP you need to teach. Use the brief on this page, save the first output, and log the credits for one correction.
Create the module in Golpo · Compare plans

Frequently asked questions

Which AI keeps every required SOP step?

A controlled same-source test is how you answer that for your procedure. Score the first Golpo output for missing, reordered, altered, and invented steps before you consider production speed. This page does not publish scores for other products.

Can the result be called compliant?

A generated video is not automatically compliant. Compliance depends on the approved source, the final media, the delivery system, the jurisdiction, accessibility, and qualified review.

How should a changed policy update the video?

Version the source, identify the affected scenes, regenerate or edit, review again, replace the published copy, and keep the prior approved version for audit. Record the selected duration and the credits charged for that new version.

What should an L&D reviewer approve?

The learning outcome, required steps, warning, exception, escalation owner, assessment alignment, accessibility package, final video, and effective date.

Why not compare marketing demos?

A marketing demo uses a friendly script. Your test uses one approved SOP and records the work and credits required to reach an acceptable version.

Is the NIST video the same-SOP test?

No. The embedded video is a Golpo demonstration made from NIST CSWP 29, the Cybersecurity Framework 2.0 PDF. That document is a public framework, not an internal procedure with a trigger, a warning, an exception, and an escalation owner. Use the SOP pack on this page for the controlled test.