Most AI video tools have one stage: you write a prompt, they decide everything else, you get back a locked render. RizzGen has ten, and you sign off on each before the next begins. This page is the whole thing, start to finish, with nothing hidden behind a signup.
Everything before generation is free. Research, concept, script, and unlimited revisions cost nothing. Credits apply when a voice, image, music track, video clip, or export is generated.
One project, two surfaces: the conversation with Rizzi and the full editor, sharing the same state.
Jump to any stage.
Every video project begins with a conversation, and every conversation is tied to a video project.
The primary interface is Rizzi, RizzGen's creative agent, reached through a conversational chat. The secondary interface is the full editor, a scene-level production environment. They are not separate products. They share the same project context, and you can move between them at any point without losing state.
The left sidebar holds New Chat, Projects, Library, Context, and Connectors. Every chat session in that sidebar is a video project. Past chats are past productions.
A new chat opens on a single input:
Describe your idea — I'll craft the script, visuals, and voice for you.
Below it sit Example Videos: curated, completed sessions you can play in full before you sign in. The input bar has a + button for attachments and a Quick mode selector for controlling generation speed and cost.
Rizzi is not a production executor waiting on a finished brief. It's a creative partner that can help you build one.
You can arrive with anything:
For partial or open inputs, Rizzi can research what's currently performing on the platform you named, suggest multiple creative angles on the same topic, propose aesthetic references matched to your established style if a Context is loaded, ask clarifying questions to narrow the direction, and identify the emotional or narrative logic that would make the topic land with your audience.
This upstream creative work is the stage every other AI video tool skips. Specific intent produces specific output. Vague prompts produce generic output. Rizzi's job here is to get you to something specific before a single credit is committed.
Rizzi researching a topic and returning angles before any production begins.
A written plan, in the chat, that you approve before anything is generated.
Once a direction exists, Rizzi develops it into a structured concept plan, shown as an inline panel in the conversation. It covers:
You read it and respond. If the tone is off, the visual approach doesn't match your brand, or the angle isn't quite right, you say so and Rizzi revises. That loop runs as long as you need it to, and it costs nothing.
Nothing is generated before this point. You have full visibility into and control over the creative direction before any production resource is committed.
The concept plan arriving inline, ready to be revised or approved.
Concept approved. Two ways forward, and you can change your mind at any time.
The concept loads into the full editor with the production plan as the organising structure. You take direct control of every production decision. Rizzi stays available as a conversational assistant in the chat panel.
For: creators with a clear vision who want maximum precision from the start.
Rizzi executes the production plan step by step, presenting each component for approval before moving on. You stay in the conversation, reviewing and directing at each checkpoint.
For: creators who want Rizzi to carry the production load while they keep approval authority at every stage.
Both paths are reversible. Start in chat, decide you want granular control, open the editor. Work in the editor, return to chat to make a change conversationally. The project context is shared across both surfaces, so nothing is lost in either direction.
Rizzi orchestrating a production plan step by step in chat.
Not a flat document. A production blueprint.
The script is generated against the approved concept plan, structured scene by scene, with narration, on-screen text notes, and pacing cues embedded.
From here you can accept it and proceed, request changes to tone, vocabulary, pacing or structure, rewrite sections directly, ask Rizzi for alternative versions of specific lines or segments, or ask for the whole thing shorter, longer, more or less formal, or in a different voice.
The script is not locked until you approve it. Catching a script problem here costs nothing. Catching it after visuals are generated costs credits and time.
Two pathways: pick a voice, or bring your own.
Rizzi presents voice options matched to the concept's tone, platform, and target audience. The voice selector appears as a dynamic component inline in the chat — not a settings page, not a separate workflow. You select, review a sample, and approve. The voiceover is generated against the approved script and placed in the project timeline.
Upload your own voice recordings and RizzGen generates voiceover in that voice. This matters most for creators whose personal brand is built on their voice, and for channels where consistency across every video is part of the identity.
Selecting and previewing a voice inline, without leaving the conversation.
Scene-level, not video-level. This is the difference.
You don't generate one video from one prompt and receive a complete output. You direct each scene individually, with full control over what gets generated for each segment of the timeline.
RizzGen routes across multiple video and image generation models — Seedance 2.0, Kling, Veo, and others. Each has different strengths for different content. You can select the model for a scene yourself, or let it route automatically based on what the scene needs.
Routing is not random. The platform selects on the basis of the requirement: cinematic motion, animation style, product accuracy, character consistency.
For every scene you can specify:
Upload reference images of characters, objects, products, or visual elements that need to appear consistently across multiple scenes. Generation is conditioned against those references to hold visual consistency across the timeline.
This addresses one of the most persistent failures in AI video: a character or product that looks different in every scene, so the credits you spent produced footage you can't cut together.
For video clip generation you can specify a start frame — a particular image or frame the clip should begin from. That gives precise control over how a scene opens and keeps continuity between adjacent clips.
If a generated clip is wrong, you regenerate that scene. Not the video. Only the scene that needs to change is regenerated, and everything else stays intact.
This is what makes professional-grade output achievable without unlimited credit spend. You don't need every clip to be right on the first generation. You need to be able to fix exactly the thing that's wrong without disturbing what's already right.
Directing a scene: style, camera, references, and framing set before the clip is generated.
Conditioned against the tone you already established.
Music generation is prompt-driven and conditioned against the project's established creative direction. If your concept was built around "cinematic melancholy, slow-burn pacing, minimal instrumentation," you prompt for music against those parameters rather than starting from a blank field.
AI-generated sound effects can be placed at specific moments in the timeline.
Upload licensed music, brand audio assets, or sound design you've already produced. Uploaded audio sits in the project timeline alongside generated audio.
Available at any point — before, during, or after any production stage.
The timeline shows the full video as a sequence of scenes, each scene a clip in the visual track. The voiceover waveform sits above the scene track; music can be added below. Scenes are rearranged by dragging. The scrubber shows playhead position and current timestamp.
From the editor you can regenerate any individual scene with adjusted parameters, replace a clip with a different generation, adjust the start frame of any clip, add, remove, or reorder scenes, open any scene to change its visual direction, add on-screen text, and adjust captions.
Quick Edit is for fast, high-level adjustments — overall tone, pacing, or style, without going scene by scene. Advanced Edit opens every per-scene control and parameter for maximum precision.
The chat panel is reachable from inside the editor at any point. Ask Rizzi to make a change conversationally — "change the third scene to use cooler lighting" — and it executes against the existing project state.
The editor: preview, timeline, per-scene panels, and the script viewer.
RizzGen's persistent creative memory. The reason output gets more specific to you over time instead of drifting back to generic.
A Context is a structured profile built through conversation with Rizzi. Each one can contain:
Conversationally. You tell Rizzi what the context is for and describe the brand or channel. Rizzi asks clarifying questions, expands what you've said into structured reference files, and hands back editable files you can open and add to directly for anything the conversation missed.
In any chat session, type /ContextName to load one. Rizzi enters the session already
knowing the brand, the aesthetic, the voice, and the platform, and every generation in that
session is conditioned against those parameters. If you run multiple channels, multiple clients,
or multiple formats, you keep a separate context for each and switch with a single command.
This is the answer to the thing that makes AI video disappointing after session one: the model starting from zero every time and regressing to its own average. A creator who has built out their context isn't starting from zero. Rizzi knows what their brand looks like, sounds like, and how it communicates, and the output reflects that accumulated identity.
Nothing you generate is stranded in the session that made it.
The Projects section in the left sidebar organises all your work. Each project holds its chat session and its generated video together, so reopening a project reopens the reasoning behind it, not just the file.
The Library holds every generated media asset across all sessions — video clips, images, voiceovers, music. All of it is available for reuse in other projects.
Export in the aspect ratio and format your platform needs. The timeline is format-agnostic: 16:9 cinematic, 9:16 vertical, and square configurations are all supported.
The ten stages are the full path. You don't have to walk all of it every time.
Start with anything, vague or specific. Rizzi develops it into a concept, then executes scripting, voice, and visuals step by step with your approval at each stage, and delivers a finished video.
Best for: creators who want Rizzi to carry production while they direct from the conversation.
Develop the idea into a concept with Rizzi, approve it, then open the full editor and build the video yourself with scene-level precision.
Best for: creators who want to work out the creative direction conversationally but handle visual production themselves.
Skip the conversation. Open the editor and build scene by scene. Every editor control is available from the start.
Best for: experienced creators with a fully formed production plan who want to execute immediately.
Browse the Showcase, find a video whose style, format, or aesthetic matches where you're headed, and hit Create Similar to open RizzGen with those creative parameters pre-loaded.
Best for: creators working from an example they want to use as a springboard.
Load a Context with /ContextName before you begin. Every generation in the
session is conditioned against the brand, aesthetic, and voice encoded in it.
Best for: established brands and agencies who need consistent output across many videos without re-explaining their identity each time.
Reopen any past project from chat history or Projects and continue the conversation. Change specific scenes, regenerate with different parameters, add new scenes, revise the script.
Best for: iterative production, client revision rounds, or extending a series with consistent direction.
Six of the ten stages are free. Here's the exact line.
RizzGen runs on credits, not tiers. You get the full studio from day one and buy credits when you need them. There's no plan to upgrade to in order to unlock a model or a resolution.
Credits you buy today are usable six months from now with no penalty. Creative work doesn't run on a monthly billing schedule, and you shouldn't lose what you paid for during a quiet month or a busy one.
The reason the workflow front-loads six free stages is arithmetic. A concept problem caught in the chat costs nothing. The same problem caught after twelve clips have rendered costs twelve clips. Every approval gate exists to move the cost of being wrong to the cheapest possible point.
| Stage | One-click generators | RizzGen |
|---|---|---|
| Before the brief exists | Not supported. You need a prompt to start. | Rizzi researches and develops the direction with you. |
| Concept | Decided by the model, invisible to you. | Written out, revised, and approved before anything renders. |
| Script | Generated inside the render, often unseen. | Scene-by-scene, editable, locked only when you say so. |
| Visual direction | Whatever the model inferred from your prompt. | Set per scene: style, camera, references, framing, start frame. |
| Model choice | Fixed, or gated behind a higher plan. | Per scene, chosen by you or routed to the scene's requirement. |
| Fixing one bad shot | Regenerate the whole video, pay again. | Regenerate that scene. The rest of the timeline is untouched. |
| Session two | Starts from zero. Output drifts back to the model's average. | Load your Context. Rizzi already knows the brand, voice, and look. |
| Billing | Monthly tiers, "unlimited" with caps inside it, expiring credits. | Credits that never expire. No tiers. Full studio from day one. |
Longer form on this: Scene-based vs prompt-to-video, what directorial control actually means, and why AI video gets generic after session one.
No. RizzGen is built to start before the brief exists. Open a chat with an open question — "what kind of video should I make for my skincare brand this month?" — and Rizzi researches what's performing on your platform, proposes angles, and asks clarifying questions until the direction is specific enough to produce against. That upstream stage is the one most AI video tools skip, and it's why their output is generic.
The free stages take as long as you want them to — the conversation isn't metered. Generation is the part with a clock on it: individual scene renders typically complete in under three minutes depending on the model, resolution, and frame rate, and rendering runs in background cloud pipelines so your editing flow isn't blocked while it works.
Yes. The ten stages are the full path, not a mandatory sequence. You can open the editor directly and build scene by scene without any conversation, or start from a Showcase video with parameters pre-loaded. See the six entry points above.
No. Visual production is scene-level, so you regenerate the specific scene that's wrong and the rest of the timeline stays exactly as it is. You don't need every clip right on the first generation — you need to be able to fix the one thing that's wrong without disturbing what's already right.
At any point, in either direction. They're two surfaces onto the same project and share the same state. Start in chat and open the editor when you want frame-level control; work in the editor and return to chat to say "change the third scene to cooler lighting." Nothing is lost either way.
No prior editing-software experience is required — Rizzi handles technical execution through conversation. What the workflow does require is a creative point of view: you develop the concept, direct the script, and specify visual direction scene by scene. When you want frame-level precision, the full editor is there, tied to the same project.
Yes. RizzGen supports hybrid workflows: import your own video clips, image assets, reference storyboards, audio tracks, and voiceovers directly into the timeline alongside generated material.
Your contexts are editable reference files you can open and read at any time. They're built through conversation but they aren't locked inside it.
Yes. The Showcase is fully playable without an account, and Example Videos on the new-chat screen are completed sessions you can open and read before you sign in. Chat, concept, and scripting are free once you're in, so you can run the whole upstream workflow before spending anything.
Open a chat, describe something you've been meaning to make, and see what Rizzi comes back with. Nothing is generated and nothing is charged until you approve the plan.