Island's Scenic AI Control Plane Launch Film

A narrated 16:9 B2B launch film makes a new category feel suddenly widespread, turns that scale into operational risk, then presents a product as the governing layer. It moves from bare kinetic typography and abstract nodes into sunny, rolling landscape composites with translucent, close-up UI proofs, then ends with a concise benefit and logo.

Why it works

• The first promise is legible with sound off: a three-word typed opening turns into a giant category word by 0:01, followed by a field of repeated dots at 0:04–0:06.
• The 0:08–0:34 problem section names different failures while its long shot changes internally at 0:08, 0:22, 0:25, and 0:30; the film creates progression without cutting every sentence.
• The 0:43–1:30 proof sequence shows actual UI categories, access toggles, a blocked action, audit detail, and spend data. Each product claim gets a corresponding visual instead of a generic beauty shot.

Tools you'll need: Remotion for motion graphics · an image model (GPT Image, Midjourney) · screen recordings of the product · ElevenLabs or a recorded voice · Claude alone: partly

Hook formula: visual pattern interrupt + category call-out: “Let's talk about {{NEW_CATEGORY}}.” Then show it multiplying before naming {{CONSEQUENCE}}.

Why it holds: The first frame is almost empty, the next is oversized, and the first sentence leaves over two seconds of space. That change in scale gives the viewer a simple subject before the dense explanation begins.

Retention map: re-hooks at 0:08 (problem turn), 0:34 (landscape reveal), 0:43 (first real UI), 1:03 (assistant scene), and 1:19 (spend dashboard) · payoff at 1:31–1:38 · logo CTA at 1:39–1:43 · hard stop, no loop.

Mix: Approximately 35% white-field type and abstract graphics, 50% landscape/UI composites, 15% scenic title and dark end cards (estimated from sheets). No talking head or live presenter.

Rhythm: 25 definite cuts in 103 s, but 19 new text/graphic states inside shots. The unusually long shot 5 runs 27.36 s and advances through at least four internal states; it should be storyboarded as a sequence, not held as one static image. Product claims from 0:43 onward generally receive a new UI view or close-up within 2–5 seconds.

Editor rules (inferred): Move from a wide scenic UI card to its proof detail; keep UI legible at the moment the narrator names it; use scale changes and dissolves inside longer scenes; reserve dark green full-frame typography for the final promise.

Full shot list

• Look: Two worlds: warm off-white #F3F1EE for ideas, then open daylight landscapes with sky blue #B8D5DC, grassy green #6F8D5D, deep pine #04231F, charcoal text #343334. Product UI is frosted white with soft shadows and rounded corners. Product proof badges add pale red #F1B8C1, yellow #F6D96D, and mint #BCE8D4. Colors estimated from the sheets.
• Typography: Closest free sans is Inter Regular/Medium for narration-like text and UI; Inter Light for scenic titles. Use Libre Baskerville Italic only for an occasional editorial emphasis in the final promise. Opening type is roughly 3% of frame height, then the category word exceeds 20%; UI table headers are about 2–2.5% of height. No stroke or caption box.
• Captions: None in the usual subtitle sense. Use short center headlines or selective kinetic words, never duplicate every spoken word. Show a key term on the utterance, then give the eye a pause.
• Graphics: Gray/gray-lilac dot arrays, pastel gradient sphere, thin orbital lines, concentric translucent rings, discrete UI callouts. Keep object motion smooth: 6–12 frame eases for entrances, slow drifting/push-ins for scenic cards, instant cut or short dissolve for subject changes.
• B-roll: Dreamlike but photographic open hills, trees, mountains, water, and sky; wide landscape plates are backgrounds for the UI rather than a subject in themselves. Hold a plate 2–6 s, move its interface layer independently.
• Transitions: Mostly hard cuts between distinct scenes, with short internal dissolves/zooms on UI and type. A few tight punch-ins make small labels legible. No recurring whip or glitch system is visible.
• Branding: Branded proposition appears at 0:34; small website/logo closes 1:39–1:43. For a new product, put {{PRODUCT}} on the scenic reveal and {{URL}} on the final solid card.

Measured: 242 words; 147 wpm over the full film / 197 wpm during voiced passages · 30 pauses of at least 0.25 s, median 0.66 s, longest 2.4 s before the enforcement explanation at 0:59.6 · median pitch 206 Hz with highly animated 5.7-semitone spread · 13 falling, four rising, seven level sentence ends. Emphasis lands on the ubiquity claim at 0:03.7, the negations around 0:08–0:13, and proof words around 0:39, 1:03, 1:09, and 1:23.

Described: A clear, upbeat commercial narrator in a higher register. The tone begins conversational, builds urgency through short negative sentences, then becomes confident and explanatory for the product. Gender and accent are not essential to the format.

Voice direction: Read as a smart, approachable product narrator addressing a business decision-maker. Begin with curiosity and a full beat of space after naming {{CATEGORY}}. Make the risk list clipped and specific; slow slightly at the question and product reveal. Land capability nouns cleanly so each UI state can appear on the word. End the benefit with calm certainty, then leave room for the logo.

• Music: A continuous tonal, music-like bed is measured at about −27.9 dBFS, roughly 9.5 dB below the voice; a restrained ambient/electronic character is inferred rather than verified by the measurement. The bed sustains the hook pause and fades down after the last word (−45.4 dBFS).
• Effects: The analyzer found louder accents on 36% of cuts, but no clear evidence for a consistent SFX-on-every-cut pattern. If adding effects to a new film, keep them subtle: one soft UI tick per key state, one lift into the reveal, and no loud whoosh on routine table zooms.
• Silence and space: The 2.14 s gap after the category word and the 2.4 s gap before the 1:02 explanation are structural. Let graphics and the bed carry them.
• Mix: Reference −14.4 LUFS integrated; measured true peak +0.2 dBFS. For a new export, target around −14 LUFS with a limiter at or below −1 dBTP to avoid codec clipping.

• Title formula: Product/category announcement or work sample: “How {{PRODUCT}} brings {{CATEGORY}} under control.” The posted caption instead describes the editor's commission and invites DMs; that is portfolio packaging, separate from the film's narrative.
• Cover: A large frosted white AI skills table centered over green hills and pale sky. The table takes about 70% of frame width; its colored risk labels are the immediate visual hook. For your own film, cover with the most specific UI proof, {{HERO_SCREEN}}, over a clean scenic plate.
• Description and hashtags: Source post is a brief editor work sample with a direct-message CTA and link; no hashtag or pinned-comment pattern is supplied.
• Audience signals: At capture, 1,368 views, 45 likes, four comments, and two reposts (32.9 likes and 2.92 comments per 1,000 views). No most-replayed data or comment text was supplied, so particular reactions cannot be assigned to a beat.

The budgets below total about 240 spoken words; allow pauses between sections. Replace every claim with a true feature and visual proof.

[0:00–0:08 · CATEGORY · ~17 words]
Let's talk about {{CATEGORY}}. {{SHORT_UBIQUITY_CLAIM}}.

[0:08–0:27 · STAKES · ~44 words]
{{CATEGORY}} can do {{USEFUL_WORK}}. But {{AUDIENCE}} still has to manage {{RISK_1}}, {{RISK_2}}, and {{RISK_3}}. Give each risk one short sentence and one graphic.

[0:27–0:34 · TURN · ~15 words]
What if {{AUDIENCE}} could {{DESIRED_CONTROL}}? Introduce an affirmative answer without explaining it yet.

[0:34–0:42 · REVEAL · ~19 words]
Meet {{PRODUCT}}. State one plain-language category proposition. Show the whole system.

[0:42–0:51 · PROOF 1 · ~22 words]
Show {{VISIBILITY_FEATURE}} in the real UI. Name the hidden or risky case it catches.

[0:51–1:03 · PROOF 2 · ~27 words]
Show {{ACCESS_FEATURE}} with a before/after state: identity, permission, then outcome.

[1:03–1:19 · PROOF 3 · ~37 words]
Stage a specific {{UNSAFE_ACTION}}. Show {{PRODUCT}} intervening and creating {{AUDIT_EVIDENCE}}.

[1:19–1:31 · PROOF 4 · ~28 words]
Show {{METRIC}} and {{COST_OR_PERFORMANCE_DECISION}} in a legible dashboard.

[1:31–1:43 · BENEFIT/CLOSE · ~28 words]
Translate all four capabilities into one human outcome for {{PEOPLE}}. Close on {{PRODUCT}} and {{URL}}.

Toolkit

• Motion graphics: Remotion. Builds typography, risk diagrams, and glass UI
• Generated stills: an image model (GPT Image, Midjourney). Creates the daylight landscape backgrounds
• Screen capture: screen recordings of the product. Shows authentic interface proof
• Voice: ElevenLabs or a recorded voice. Carries the category and risk story

Claude can make the motion and the edit; the parts above that need other tools are the rest. The landscape composites and product proof need the most care.

Script prompt

Write an original 100–105 second B2B product-film narration for {{PRODUCT}} about {{TOPIC}}, aimed at {{AUDIENCE}}. Use the nine timed beats and word budgets in section 10. Open with a short category call-out, build three concrete operational stakes, ask one turning question, then show four true product capabilities in a causal order: visibility, access, intervention/audit, and optimization. For every claim, specify an actual UI state or graphic the editor can show. Keep the tone conversational, incisive, and optimistic; about 240 spoken words with room for the hook and reveal pauses. Do not reuse the reference's script, product names, or brand claims. Return a timecoded two-column voice/visual draft plus a list of assertions requiring product verification.

Voice prompt

One lively, warm commercial narrator with a clear higher-mid register. Speak quickly in the proof sections but give the category word and product reveal deliberate space. Crisp consonants, varied pitch, short decisive sentence endings, no exaggerated hype. Use an original voice; do not imitate any person.

Visual prompts

• Scenic plate: Wide 16:9 softly sunlit rolling green hills, sparse trees, distant blue mountains, pale cyan sky, spacious center for a floating interface, realistic landscape photography, gentle camera drift, no logos, no readable text.
• Abstract problem: Warm off-white canvas, charcoal sans-serif words loosely arranged, charcoal and dusty lilac nodes linked by thin lines, one soft pastel gradient orb, minimalist enterprise motion graphic, no product logos.
• Product UI composite: Rounded frosted white enterprise dashboard floating in front of open landscape, subtle glass edge and natural shadow, pale status chips in mint yellow and rose, clean readable columns; use supplied {{PRODUCT}} UI captures as the source of all labels and values.

Caption spec: No full subtitles in the reference. For accessibility, add optional separate captions in Inter Medium, white on a semi-opaque deep-green lower strip, two lines maximum, without covering the key UI; keep selective title words in the scene itself.

Music prompt: Restrained bright ambient electronic bed, light pulse, no lyrics, low enough to keep dense narration clear, brief lift at 0:34 and resolved tail at 1:39. This character is a production choice inferred from the measured tonal bed.

1. Verify {{PRODUCT}} claims and capture each required UI state at display resolution before animation.
2. Write and record about 240 original words to 103 s, protecting the opening gap and the reveal pause.
3. Make two visual kits: minimal off-white typography/nodes and the consistent scenic glass-UI world.
4. Build the nine beats, then the 26-shot timeline. Treat the 27 s problem shot as four or more animated sub-scenes.
5. Animate UI from wide context to proof detail; match each spoken capability to one readable change in the interface.
6. Add a continuous low bed and sparse state sounds; mix around −14 LUFS, peak-limit to −1 dBTP.
7. Export 16:9 at 25 fps and make a cover from the clearest product UI moment.

• Category is understandable by 0:02; the risk is clear by 0:08.
• Roughly 26 shots and 44 total picture changes over 103 s; the long problem passage has internal motion.
• All four proof claims have legible UI evidence: asset/risk, access state, blocked action/audit, and metric.
• Narration is near 240 words, around 147 wpm over runtime, with the two deliberate pauses preserved.
• Scenic cards share the same palette and perspective; UI remains readable on a phone.
• No borrowed brand marks, interface values, scenery, script, music, or voice.
• Audio approaches −14 LUFS integrated and stays below −1 dBTP.
• The final benefit follows from the proof, with a clean four-second logo/URL hold.

• Source signature: Island's name, its AI-agent control-plane claims, product interface, specific green landscape treatment, risk table, and editor's portfolio caption. Do not transplant these assets or exact assertions.
• Transferable format: Category enlargement → accumulated operational stakes → scenic system reveal → a chain of UI proofs → human benefit → quiet brand hold. The strongest reusable move is matching each abstract claim to a concrete interface state at a predictable point in the narration.
• Use your own product UI, verified numbers, licensed or original scenic plates, an original narrator, and original music.

Source: https://x.com/Shajareditx/status/2113843002086949354 · measured with dissect 1.1.0 on 2026-10-26 · visual estimates and aesthetic sound descriptions are marked or described as inferences