How to Make Kurzgesagt-Style Videos with AI
Last updated: March 2026 — 9 min read
Quick answer
You can create Kurzgesagt-style science explainer videos using AI tools like Onira - combining AI-generated visuals, professional narration, and cinematic pacing to produce documentary-quality explainers from a text prompt. AI gets you 70–80% of the way. The visual consistency, narration quality, and pacing are all achievable. The custom flat-design illustration style is not yet replicable.
Example: Kurzgesagt-Style Science Video
AI-generated science explainer visuals from Onira
What Makes Kurzgesagt's Style
Kurzgesagt - In a Nutshell has 24 million subscribers and averages hundreds of millions of views per video. That success comes from a very specific formula that took years and a team of 40 people to develop. Understanding the formula is the first step to approximating it with AI.
Clean, purposeful visuals
Every frame in a Kurzgesagt video serves the explanation. Visuals are not decoration - they are the argument made visible. Complex processes are broken into simple, sequential images. The flat-design aesthetic removes visual noise and forces clarity. Nothing is on screen unless it is earning its place.
Accessible science communication
Kurzgesagt takes genuinely complex topics - quantum entanglement, stellar nucleosynthesis, existential risk - and explains them without dumbing them down. The writing is precise. Analogies are well-chosen. The audience leaves understanding something real, not a simplified cartoon of it.
Professional narration
The voice is warm, measured, and authoritative without being cold. Pacing is deliberate - key ideas are given space to land. There are pauses. The narration does not race through the script. It treats each concept as worth understanding, not just consuming.
Consistent color palette
Each Kurzgesagt video has a defined color world: a dominant hue, a consistent palette of supporting tones, and carefully controlled contrast. This visual consistency makes the video feel cohesive even when scenes change rapidly. It is the equivalent of a unified visual register in live-action film.
Smooth, intentional pacing
The editing rhythm matches the emotional arc of the content. Information-dense passages use faster cuts and more visual elements. Contemplative moments use wider shots and longer takes. The pacing is felt more than noticed - a viewer does not feel rushed or bored.
Can AI Replicate This Style?
Honest assessment: AI can get 70–80% of the way toward a Kurzgesagt-quality science explainer. The gap is real and worth understanding before you start.
What AI can match
- Narrative structure
AI script engines understand documentary storytelling: hook, context, rising complexity, resolution. Gemini 3.1 Pro writes 60–80 scene scripts that hold together as arguments.
- Narration quality
ElevenLabs voices are now indistinguishable from human narrators for most listeners. Tone, pacing, and warmth can be configured per project.
- Color consistency
consistent visual treatment applied uniformly across all AI-generated footage creates a cohesive visual identity that mirrors Kurzgesagt's palette control.
- Cinematic pacing
Automated editing systems can vary shot length based on information density and emotional weight - the same logic a human editor applies.
What AI cannot match (yet)
- Custom illustration style
Kurzgesagt's flat-design characters and assets are built and refined over years. AI cannot generate the same consistent character design across 500 scenes.
- Frame-by-frame animation
Kurzgesagt uses motion graphics artists who animate objects in specific, purposeful ways. AI video generation produces motion, not directed animation.
- Visual metaphor depth
The most memorable Kurzgesagt moments are when a visual metaphor crystallizes an abstract concept. AI visuals are descriptive; they rarely achieve that level of conceptual elegance.
The practical implication: AI can produce a science explainer that looks and feels professionally made and genuinely educational. It will not look like a Kurzgesagt video if you put them side-by-side. But it will look far better than anything a solo creator could produce manually - and it will share the DNA of what makes that format work.
For YouTube, that is often enough. A channel that consistently produces well-narrated, visually coherent science explainer videos at quality will build an audience regardless of whether it uses flat-design illustration or AI-generated footage.
Step-by-Step with Onira
Here is the practical workflow for producing a Kurzgesagt-style science explainer using Onira's end-to-end production pipeline.
Choose your topic and angle
Pick a science topic with genuine depth and broad appeal. The best topics have an element of wonder or surprise - something that makes a viewer think "I never thought about it that way." Avoid topics that require on-location filming or real interviews. Biology, physics, space, history of science, and speculative scenarios all work well. Be specific about the angle: not "black holes" but "why black holes don't actually suck things in" - a specific claim the video will prove.
Write a detailed prompt
Give Onira's Script Engine a full brief: topic, angle, intended length (8–12 minutes works well for this format), target audience, tone, and the emotional arc you want to achieve. Specify that you want a documentary-style narration, not a lecture. Mention pacing preferences - e.g., "start with a provocative question, build to a revelatory middle section, end with a broader reflection." The more specific your prompt, the stronger the script.
AI produces the full video
Onira runs the full production pipeline: Gemini 3.1 Pro writes and structures a 60–80 scene script, ElevenLabs generates the narration audio with appropriate pacing, AI models generate visuals for each scene (routing to Kling for motion, Hailuo for atmospheric hero shots, Veo for photorealistic sequences), AI music is generated to match the emotional arc, and consistent visual treatment is applied uniformly. The result is a finished video in 10–20 minutes.
Review, refine, and publish
Watch the full video. Check the script for factual accuracy - AI can hallucinate details on niche topics. Identify any scenes where the visual does not match the narration well and regenerate those individually. Adjust the narration tone if needed. Once satisfied, export in YouTube-optimized format. Onira generates suggested titles, descriptions, and tags alongside the video.
Pro tip: Before publishing, spend 15 minutes verifying the key scientific claims in the script. AI script engines are accurate on well-documented topics but can introduce errors on niche or recent research. Your credibility as a science channel depends on getting the facts right - that investment compounds over time.
Example Topics That Work
These topics share the characteristics that make the Kurzgesagt-style format shine: they are visually compelling, scientifically rich, and impossible to film in real life. All of them are well-suited to AI-produced educational video.
The Scale of the Universe
From quarks to the observable universe - a visual journey through 40 orders of magnitude. Consistently one of the most-watched science formats on YouTube.
How the Human Immune System Works
Cell-level biology is invisible in real life and perfect for AI visualization. Complex enough to deserve a 10-minute treatment, accessible enough for a general audience.
What Would Happen If a Neutron Star Hit Earth
Speculative physics scenarios are a Kurzgesagt staple. They combine awe-inspiring visuals with genuine science communication. The drama writes itself.
The History of Climate Change Science
A historical narrative about how humans understood climate - from Arrhenius in 1896 to the IPCC today. Data visualization + historical recreations play perfectly to AI strengths.
How Vaccines Train the Immune System
Molecular biology visualized: antigens, B-cells, antibodies, and memory cells. The Kurzgesagt immune system videos have hundreds of millions of combined views. The demand is proven.
The Fermi Paradox Explained
One of the most-discussed concepts in science communication. Abstract ideas about civilizations and probability translated into vivid visuals - exactly what AI handles well.
How the Internet Actually Works
Infrastructure and protocol explainers are underserved. Routers, data packets, submarine cables, server farms - all visually compelling when rendered with AI.
The Birth and Death of Stars
Stellar evolution from nebula to white dwarf or supernova. Timescales of billions of years, rendered in 12 minutes. One of the easiest topics to make look cinematic.
Notice the pattern: each of these topics involves processes, scales, or scenarios that cannot be filmed. They require visualization to understand. AI-generated footage - far more than stock footage - can actually show what is being described, rather than approximating it with tangentially relevant clips.
This is the core advantage of AI-generated visuals over stock footage for science content. When your script says "the neutron star collapses into a singularity," AI can generate that. Stock footage cannot.
Tools You Need
The minimal tool stack for producing Kurzgesagt-style science explainers with AI.
Onira
End-to-end productionHandles the entire production pipeline: script (Gemini 3.1 Pro), narration (ElevenLabs), visuals (Kling, Hailuo, Veo routed per scene), music generation, visual finishing (cinema LUTs), and final assembly. One prompt, one finished video. A 10-minute science explainer costs approximately ~$26 and renders in 10–20 minutes.
Midjourney
Optional - custom stillsUseful if you want to develop a specific branded illustration style - custom characters, recurring visual motifs, or a defined aesthetic that differs from Onira's default output. You can generate Midjourney stills and incorporate them as visual anchors in your videos. Not required for most science explainer topics where the content itself provides enough visual interest.
That is genuinely the full stack. The tools that a solo creator previously needed - separate subscriptions for scripting, narration, stock footage, music licensing, and editing software - are all replaced by Onira's pipeline. The total monthly cost for producing two science explainers per week is approximately $149–$349 depending on volume, compared to $150–$300+ for assembling individual tools.
The more important saving is time. Traditional production of a single 10-minute science explainer with a multi-tool stack takes 12–20 hours. With Onira, it takes 1–2 hours including review and refinement. That time difference is what makes a consistent two-videos-per-week publishing schedule achievable as a solo creator.
Frequently Asked Questions
Can AI fully replicate Kurzgesagt's animation style?
Not yet. Kurzgesagt uses a distinctive flat-design illustration style with custom character rigs and frame-by-frame motion graphics. AI currently cannot reproduce that level of visual consistency across an entire video. What AI can match is the structural quality: clear narrative, cinematic pacing, professional narration, and a cohesive color palette. You can get 70–80% of the Kurzgesagt effect; the custom illustration layer requires human artists.
How long does it take to produce one video with AI?
With a tool like Onira, you can go from a text prompt to a finished 8–12 minute science explainer in under an hour. The pipeline handles script generation, visual production, narration, music, visual treatment, and assembly automatically. Compare that to the 6–8 weeks Kurzgesagt reportedly spends on a single video with a team of 40 people.
What topics work best for this format?
Topics that are difficult or impossible to film in real life work best - deep space phenomena, biology at the molecular level, historical civilizations, abstract physics concepts, speculative futures. If the topic benefits from visualization rather than real-world footage, AI-generated visuals are a natural fit and often look better than stock footage alternatives.
Do I need Midjourney for custom stills?
No - Onira generates all visuals natively, routing each scene to the best AI model for that type of content. Midjourney is useful if you want to create a specific branded illustration style or match a custom aesthetic more precisely. For most science explainer topics, the native AI visuals from Onira's multi-model pipeline are sufficient.
Will YouTube monetize AI-generated science explainer videos?
Yes, as long as the content meets YouTube's quality and originality standards. AI-generated content is explicitly allowed under YouTube's current policies. The risk is with fully automated, low-effort content - a well-crafted science explainer with original narrative framing, accurate information, and quality production has no monetization issues. Channels like this have been building audiences successfully since 2025.
Build your science channel with AI
Onira produces cinema-quality science explainers from a single prompt - professional narration, AI-generated visuals, and cinematic visual finishing included.
From $149/mo · Cancel anytime