Creating Video with Antigravity + Remotion

I created this video using Google Antigravity and Remotion using remotion skills. I used Gemini Pro 3 Low.

I used this one-shot prompt:

Create a 45‑second 16:9 animated explainer video for a B2B SaaS automation platform called ‘FlowPilot’. The video should have 5 clearly defined scenes, each with its own visual composition and transitions, but a consistent design language (flat, modern illustrations, clean UI mockups, brand colors: dark blue, teal, white, subtle gradients).

Scene 1 – The pain (0–8s)

Wide shot of a cluttered workspace: multiple monitors showing different apps (email, CRM, spreadsheets, chat).
Notifications pop up rapidly, overlapping and shaking slightly to convey stress.
A silhouetted character at the desk rubs their temples; subtle camera shake emphasizes overwhelm.
On‑screen headline text: ‘Managing your operations shouldn’t feel like this.’
Background color: slightly desaturated, with a subtle vignette to feel heavy.
Transition: Notifications freeze, then shatter into small geometric pieces that fly off‑screen, revealing the next scene underneath.

Scene 2 – Introducing FlowPilot (8–16s)

Clean, bright background (white with soft teal gradient corners).
The FlowPilot logo animates in from the center, building from simple shapes that snap together with springy motion.
Thin, glowing lines extend from the logo to icons representing tools: email, CRM, billing, project management.
Each icon ‘lights up’ when connected, and a small label appears under each (e.g., ‘Email’, ‘CRM’, ‘Billing’, ‘PM’).
On‑screen text: ‘FlowPilot connects your entire stack in minutes.’
Transition: Camera dolly in towards the logo until it fills the frame, then cross‑fade into the dashboard.

Scene 3 – How it works: visual workflow (16–28s)

Top‑down view of a visual workflow canvas. Nodes represent steps like ‘New lead captured’, ‘Qualify with AI’, ‘Create deal in CRM’, ‘Send intro email’.
One by one, nodes fade in and slide into position, connected by animated arrows with directional motion.
A highlighted ‘token’ representing a lead travels along the path, triggering each step. When it passes a node, that node briefly glows.
Small text labels near each node explain the step in a few words.
On‑screen caption at bottom: ‘Build automations with drag‑and‑drop blocks.’
Transition: The camera follows the token as it exits the last node, then the token morphs into a mini dashboard icon, leading us into the next scene.

Scene 4 – Outcomes and metrics (28–38s)

Full‑screen dashboard view with 3–4 key widgets: ‘Time saved’, ‘Revenue generated’, ‘Tasks automated per week’.
Charts animate from zero to their final values with smooth easing. Percentage labels count up (e.g., ‘+63% faster response time’).
Background is a slightly blurred office environment to keep focus on the UI.
Small avatars of team members slide in along the bottom, each with a checkmark and a short benefit (e.g., ‘Ops: Fewer manual tasks’, ‘Sales: Faster follow‑up’).
On‑screen text at top: ‘See the impact across your team.’
Transition: Widgets slide aside to create space in the center, which becomes a white rectangle that expands into the final CTA screen.

Scene 5 – Call to action (38–45s)

Clean, minimal layout with logo on the left and a concise value prop on the right: ‘Automate the busywork. Scale what matters.’
Below the text, a primary CTA button animates in with a gentle bounce: ‘Book a 15‑minute demo’.
Secondary text under the button: ‘No code. No engineers. Go live in days.’
Background shows a very subtle looping animation of nodes and lines in the distance (low contrast, almost watermark‑like).
End with a 1‑second hold on the final frame for thumbnails.

Audio: Use a modern, upbeat but non‑intrusive background track. Start softer in Scene 1, then increase energy slightly when FlowPilot is introduced. Keep sound effects minimal: soft whooshes for transitions, a gentle chime when metrics increase. No voiceover for now; the video must be readable with text alone.

I don't have the audio working yet. I could improve the video with further prompting, but I wanted to see what I could get in one shot.

Combining this with Gen AI images, videos and digital twins, we should be able to create very cool stuff.

What ideas do you have for videos that can be made with this approach?

0:45

2 comments