The core finding
AI-generated video cannot render your product UI. This is a confirmed, unsolved limitation as of June 2026.
Every major AI video model — Kling 3.0 Pro, Seedance 2.0, Veo 3 — produces hallucinated, unreadable text when asked to render a software interface. Independent analysis from three separate production workflows confirms this is not a prompt quality problem. It is a structural limitation of how these models generate video. The text is garbled. The UI elements are invented. The product is not what you built.
This is a critical fact for any software founder choosing a launch video approach. If you cannot show the real product interface legibly, you are not showing your product. You are showing an AI-generated approximation of a product interface, which is worse than a raw screen recording.
"AI-generated video is confirmed unable to render legible product UI text across Kling 3.0, Seedance, and Veo 3 as of June 2026. All readable on-screen text in professional AI-assisted launch videos is added in post-production as a code overlay — never generated by the AI model itself."Kino production research, June 2026 — corroborated by Kapwing AI video comparison, Genra.ai 7-artifacts teardown, and PJ Ace's Kalshi video workflow (300+ clips generated, all UI text added post).
"A founder with access to Kling can generate an AI atmosphere clip for approximately £50 in API credits. They cannot replicate real-UI motion graphics built on a custom Remotion component library — that is the moat."Kino, 2026. The production cost per 25-second cinematic launch video is approximately £8–15 in AI generation; the value is in the motion-design expertise and the per-client component library, not the generation cost.
"The production approach used by Apple, Linear, Vercel, Stripe, and Raycast for their product launches is code-based motion graphics over real captured UI — not AI-generated video."Verified by analysis of published launch films from these companies, 2025–2026.
The comparison
Three approaches to a software product launch video.
The following table covers the five criteria that matter most for a software product launch: whether the real product interface is visible, whether UI text is legible, what it costs, how fast you get it, and whether competitors can replicate it the next day.
| Criterion | Raw screen recording |
AI-generated footage |
RecommendedReal-UI motion graphics (Remotion + code) |
|---|---|---|---|
| Shows the real product interface | Yes | No — hallucinated UI | Yes — captured at 2–3x resolution |
| Legible on-screen text | Yes | No — confirmed unsolved in 2026 | Yes — rendered in code (always legible) |
| Cinematic quality | No | Partial — atmosphere only | Yes — keynote-grade motion, real product |
| Typical cost | £0 | £50–100 DIY / £8–15k agency | £2,800/mo subscription (Kino founding rate) or £8–15k/video (traditional agency) |
| Turnaround | Immediate | Hours (DIY) / 4–6 weeks (agency) | 48–72 hours (Kino) / 4–6 weeks (agency) |
| Replicable by a competitor overnight | Yes — trivially | Yes — same tools, same cost | No — requires the per-client component library and motion-design expertise |
| Gets better with each video | No | No | Yes — Remotion component library reuse means video 5 takes 1–2h vs 8–14h for video 1 |
| Best for | Internal demos, quick context | Atmosphere B-roll only; never the product hero | Any launch where the product UI is the proof |
Verdict: for a software product launch where the product interface is what you are actually selling, real-UI motion graphics is the correct medium. AI-generated video works as atmosphere behind code-rendered UI — it should never be the hero of a product launch where legible screens matter.
The correct approach in 2026
The hybrid: code motion graphics as the spine, AI atmosphere as seasoning.
The approach used by Kino — and by every serious SaaS launch video studio — is not "AI video" or "no AI." It is a specific combination that gets the best of both:
Step 1 — Capture the real product UI at 2–3x screen resolution using Screen Studio or CleanShot. This is the foundation. No AI-generated UI, no mockups, no screenshots-of-screenshots. The real interface.
Step 2 — Build the motion-graphics spine in code (Remotion). Kinetic typography, UI overlays, animated product reveals, data visualisations. Everything that requires legible text or real interface elements lives here. Code renders crisply at any resolution.
Step 3 — Add 1–2 AI atmosphere clips (Kling, Seedance) as background seasoning only. Environments, light, hands on keyboards, abstract motion. Never product UI, never text, never the hero. Max 2 clips per video.
This hybrid is what separates a Kino launch video from both a raw screen recording and a pure AI-generated clip. The result looks like a keynote: cinematic atmosphere, real product, legible UI, motion-designed with craft. At approximately £8–15 in AI generation costs per 25-second video.
Common questions
Frequently asked.
No. As of June 2026, AI video generation models — including Kling 3.0, Seedance, and Veo 3 — cannot reliably render legible product UI text. Independent workflow analysis (Kapwing comparison, Genra.ai teardown, PJ Ace's Kalshi workflow involving over 300 generated clips) confirms that all readable on-screen text in professional AI-assisted videos is added in post-production as a code overlay, never generated by the AI model itself.
AI-generated video creates photoreal footage (environments, hands, light) but cannot render the actual product interface legibly. Motion graphics (code-based, using tools like Remotion) animate the real captured product UI with kinetic typography and overlays. For a software launch where the product itself is what you are selling, motion graphics are the correct medium — they show the real interface. AI video works as atmosphere behind code-rendered UI.
Brands like Linear, Vercel, Stripe, Apple, and Raycast use motion graphics over real captured UI because the product interface is the proof. Showing real product screens in motion is what creates launch credibility for a software product. AI-generated footage cannot capture a real interface — it generates a plausible-looking fake. Design-aware founders and technical buyers notice the difference immediately.
A DIY AI-generated clip using Kling or Runway costs approximately £50–100 in API credits. A professionally produced motion-graphics launch video from Kino costs £2,800/month on subscription (founding rate), with the underlying AI generation cost per video being approximately £8–15. Traditional agencies like Vidico or Sandwich charge £8,000–£15,000 per video on a 4–6 week timeline. The subscription model makes the professionally produced version cost-competitive for startups launching regularly.
The correct approach is a hybrid: real captured product UI (captured at 2x or 3x screen resolution using Screen Studio or CleanShot) corner-pinned into a code-based motion-graphics spine (Remotion), with 1–2 AI-generated atmosphere clips as background seasoning. This approach shows the real product, renders legible UI text, and achieves the cinematic quality that separates a keynote from a screen recording — at a fraction of the cost of a traditional agency.