Virginia Commonwealth University × Bezar Video Studio · v2 (revised per Peyton's notes)

AI Training Program — a system you own

Not a one-off video that's stale in two months — a branded, self-serve template where updating a model name or a policy is editing a line of text and hitting rebuild.

What we heard from Peyton — and what changed

No enterprise licenses. Team members get their own licenses (Claude, OpenAI, etc.) and submit to IT Governance for approval — this isn't central provisioning. We've re-framed the whole training around that reality and flagged deck lines that assumed an enterprise portal / institutional discounts to correct.
Training gates approval. Completing this training is part of getting a license approved — so it's a considerations/awareness checkpoint, not just a how-to. We'll structure it so the key "before you're approved, understand this" points land clearly.
Nebula One is coming (fall) — a single dashboard to everything in the Microsoft AI Foundry. We'll leave a placeholder module and build so it slots in when it launches.
The real frustration: "we'll have to redo this in two months." Correct — models, tools, and policies change constantly. So the deliverable can't be a frozen video.
You want to self-serve changes and new videos. This is now the center of the plan (below).

The core idea: content-as-data → video

We separate what the video says (editable content) from how it's produced (the branded template engine). Each module is a simple text/data file — the script, the on-screen bullets, the model table, the terminal steps. A one-command build turns that into a finished, captioned, VCU-branded video.

Edit a module's content file→ Rebuild (one command)→ Auto: motion graphics + VO + captions→ New video in minutes

When Claude ships a new model, you edit one row in the model table and rebuild Module 04. When policy changes, you edit that line. When Nebula One lands, you fill in the placeholder module. No re-hiring us for every change.

Why this kills the "redo it in two months" problem

Volatile stuff is data, not pixels

Model names, versions, commands, policy lines live in editable fields — change text, not animation.

Modular

Update or replace one 2–3 min module without touching the rest.

You own it

We hand over the template + a short training so your team drives it — self-serve edits and brand-new modules.

Consistent by default

Brand, voice, and layout are locked in the template, so anyone's rebuild stays on-brand.

Module map (corrected for VCU's actual setup)

#	Module	Notes / corrections
00	Start Here — what this is, and that finishing = part of approval	Frame as the governance checkpoint
01	Know Your Data (Cat 1/2/3 before you prompt)	Security priority — stays front
02	Get a License → Submit to IT Governance	Re-scoped: individual license → governance submission → approval (not enterprise portal)
03	Install with ZScaler ON	Verify commands current at build time
04	Authenticate — account login, not API keys	Drop "institutional discount" claim unless confirmed
05	Model Selection	Most volatile — model table is editable data
06	Context Management (sessions, CLAUDE.md)	Stable
07	Agentic Platforms — MCP, skills, A2A	Verify "VCU-approved MCP servers" claim w/ IT
08	Cowork & Integration Safety	Security priority
09	Cheat Sheet + Resources	Editable links/contacts
10	Nebula One (Microsoft AI Foundry dashboard)	Placeholder — build when it launches this fall

We'll confirm every technical + policy claim with your IT Governance before publish.

Look & accessibility

VCU black & gold motion graphics

Clean kinetic type, animated diagrams (data tiers, model table, MCP = "USB-C for AI").

Real terminal / UI demos

Authentic screen captures of install + OAuth + a live session.

Pro voiceover

Consistent narration; regenerated automatically when a script changes.

Captioned & 508-ready

SRT + burned-in, WCAG contrast — required for university distribution.

Phases

1 · Proof + template. Build Module 01 (Know Your Data) and the reusable template engine behind it — so you see the look and the update workflow at once.
2 · Fill the series. Script + produce Modules 00–09 from the template (content confirmed with IT Governance).
3 · Handoff + self-serve. Deliver the template, docs, and a 30-min training for your team so you can edit and create modules yourselves.
4 · Nebula One. When it launches, we (or you) fill the placeholder module from the template.

Deliverables

The module series — VCU-branded, captioned, 1080p — plus an optional stitched full version
Caption files (SRT) + burned-in versions
The self-serve template system — editable content files + one-command rebuild, handed to VCU
Written docs + a short team training on updating and creating modules
Updated on-brand cheat-sheet PDF

What we need from VCU

Confirm the re-scoped licensing/governance flow + the module map.
Brand assets (logo, exact gold/black, fonts) and voice/tone preference.
An IT Governance point-of-contact to verify technical + policy accuracy before publish.
Green light to screen-capture the tools on a VCU-managed machine.

Prepared by Bezar Video Studio. Recommended first step: Phase 1 — we build Module 01 + the template engine as a paid proof, so Peyton can watch a finished module and try editing + rebuilding it himself before we commit to the full series.