Launch films · music-cut, no voiceover · landscape 16:9
How to make a video like the ChatGPT at Work ad
The ChatGPT at Work ad works on one structural idea: white kinetic-type headline cards carry the argument, and between them the product does the work in full-bleed UI worlds. There is no voiceover — the cuts land on a music grid, and the camera moves through each seam rather than resting on it. baz ships this as the ChatGPT Work Ad skill: a ~100-second film at roughly 99 BPM across 25–27 scenes, rebuilt with your logo, your screenshots and your claims.
The shortcut: run a skill instead of starting from scratch
A skill is a context file that teaches an AI agent one specific video style — its structure, pacing, typography, motion grammar and sound design. Hand an agent the skill plus your product, and it rebuilds that video for you. The steps below are what the ChatGPT Work Ad skill already encodes, so you can follow them by hand or let the skill carry them.
baz is a library of video skills: context files that teach an AI agent a specific motion-graphics style, so it can rebuild that video for your product, with your brand, from your inputs. Run a skill inside your own coding agent — Claude Code, Codex — or directly in the baz editor.
What you need before you start
Required
- Your brand logo
- Product screenshots of the surfaces you want on screen
Makes it materially better
- A described product demo flow, so the UI worlds show a real sequence rather than disconnected screens
- Your website copy and claims, so the headline cards use your own language
- Integration partner logos for the status-ticker beat
How to make a video like the ChatGPT at Work ad: step by step
Split your argument into headline cards and product proof
This format alternates strictly: a kinetic-type card states a claim, then a full-bleed UI world proves it. Write your script as that alternation from the start — claim, proof, claim, proof. A claim with no proof beat after it has nowhere to go.
Set the beat grid before you time anything
The source runs at roughly 99.4 BPM, which is about 18 frames per beat at 30 fps. Every seam aligns to a beat boundary. Derive that grid from your own track first — if you time scenes in round seconds and add music later, nothing will land.
Keep the headline cards to one idea
These cards are white, typographic and loud. They work because they say one thing. A colour-cycling accent word gives you emphasis without a second line of copy — use that instead of adding more words.
Build the UI worlds full-bleed, not as floating screenshots
The product beats fill the frame — a desktop app, an editor, a phone. A screenshot sitting in a mockup frame reads as a slide; a full-bleed UI world reads as the product. Recreate the interface so the camera can push into it.
Move the camera through every seam
This is the signature move and the easiest thing to get wrong. The outgoing scene is mid-acceleration and the incoming scene opens mid-deceleration, so the frame appears to scale through the cut. Never let a scene reach rest exactly on a cut frame.
Land the click and the state change on the same frame
In the interactive beats, the press, the ripple and the resulting new state must all fire on one frame. A click that lands a frame early or late reads as broken, and it is the detail viewers notice without being able to name it.
What the finished video actually contains
Approximate timings for a ~100 seconds cut. Treat them as a shape to hit, not a spec — the exact durations shift with your script and your music.
| Time | Beat | What is on screen |
|---|---|---|
| Open | Logo bookend | The brand mark grows and spins on full-bleed white, accelerating into the first cut. |
| Act beats | Kinetic headline | White type cards carrying one claim each, with a colour-cycling accent word. |
| Proof beats | Product world window | Full-bleed UI — desktop app, deck editor, phone — with the camera pushing through. |
| Credibility | App status ticker | Integration marks scrolling as proof the product sits in a real stack. |
| Capability | Streaming assistant | The assistant responding live, then an output gallery scrolling through what it made. |
| Close | Logo bookend | The same mark returns, decelerating this time. The film closes where it opened. |
Mistakes that cost you the view
Resting at a seam
Do this instead: Keep the zoom mid-flight through every cut. A window that reaches its resting scale exactly on the cut frame kills the continuous-camera illusion the whole format depends on.
Linear easing anywhere
Do this instead: Accelerate outgoing, decelerate incoming. Linear motion reads as mechanical and flattens the kinetic feel instantly.
Putting the cursor in absolute frame coordinates
Do this instead: Anchor the cursor to the element it is clicking. In absolute coordinates it desyncs the moment the layout reflows.
Clicking a frame before or after the state change
Do this instead: Fire press, ripple and new state on the same frame. Off-by-one clicks are the most common tell that a UI beat was animated rather than captured.
Adding a voiceover
Do this instead: Leave it music-only. The timing contract is built on the beat grid; narration fights it and forces every seam out of alignment.
Run it from your own agent
Skills are designed to run inside a terminal agent like Claude Code or Codex, against a project you control. The same skill produces the same style whether it runs in your agent or in the baz editor.
npm install -g bazaar.it@latest
# Install this skill into your agent (Claude Code, Codex or Cursor)
baz skill add chatgpt-work-ad
baz project create --name "ChatGPT Work Ad for my product"
baz prompt "make a video like the ChatGPT at Work ad" --stream-json
baz review --summary --json
baz export start --wait --format mp4 --jsonOperations draw from a prepaid balance. See the CLI reference for the full command set.
Common questions
Does my product need to be an AI product?
No. The grammar is about alternating claims with full-bleed product proof — that works for any software with a visual interface. The AI-specific beats (streaming assistant, output gallery) can be replaced with whatever your equivalent capability beat is.
Is 100 seconds not too long for an ad?
For a launch film on your own site or a keynote, no — it is a chaptered piece with a real arc. For paid social, cut a 15–30 second version from the same project using the strongest claim-and-proof pair.
Do I need video editing experience?
No. The skill carries the editing decisions — shot order, pacing, type treatment, transitions, sound design. You supply the raw material and the facts about your product; the agent assembles the cut. You review it and ask for changes in plain language.
Can I run this in my own coding agent instead of the baz editor?
Yes. Skills are designed to be run from a terminal agent like Claude Code, Codex or Cursor. Install the CLI with `npm install -g bazaar.it@latest`, then `baz skill add <slug>` to pull the skill into your agent. The same skill produces the same style whether it runs in your agent or in the baz editor.
What does it cost?
baz bills against a prepaid dollar balance rather than a seat subscription — you top the balance up and each operation (generation, voiceover, render, export) draws a listed amount from it. There is no per-seat fee to make a video, and the price of each operation is published.
Can I change the video after it renders?
Yes, and this is the point of doing it as code rather than as a flat render. Every scene is a real component you can edit — ask the agent to re-time a beat, swap a screenshot, rewrite a line, or change the accent colour, and only that part rebuilds. You are not re-rendering from scratch each time.
Will every video from this skill look the same?
The grammar is the same — that is what makes it a skill and not a template. The content is not: your brand colours, typography, screenshots, copy and data all come from your inputs. Reusing one skill across a series is a feature, because episode three lands in the same visual language as episode one instead of being rebuilt from scratch.
Related guides
Pick a skill. Get the video.
Run the ChatGPT Work Ad skill in your own coding agent, or in the baz editor. Usage comes out of a balance you top up, with a listed price per operation — no seat subscription to make a video.