Vizzy Goes Agentic: Your AI Design Assistant Now Thinks for Itself

Until now, working with Vizzy meant picking a workflow card before you typed a single word: one for text-to-image, one for image-to-image, one for video. You had to think about the tool before you could think about the design.

That changes today. Vizzy is now agentic - it reads your intent, selects the right workflow, adjusts parameters, and executes the generation. All you do is describe what you want.

See the Vizzy overview in the docs for how agentic workflow selection works.

What "Agentic" Means in Practice

When we say Vizzy is agentic, we mean it makes decisions. Not random ones - informed decisions based on what you said, what images you uploaded, and what you've been working on in the conversation.

Here's how it works now:

• You type "a modern villa with floor-to-ceiling glass" - Vizzy recognizes there are no reference images and generates a fresh concept from your description.
• You upload a sketch and say "make this photorealistic" - Vizzy sees the image, understands the transformation request, and applies design-preserving rendering automatically.
• You say "animate the last render" - Vizzy detects video intent and generates a video without you touching a single dropdown.

No workflow picker. No settings panel. Just a conversation.

The Biggest Changes

Automatic Intent Detection

The workflow picker - those three cards at the top of every chat - is gone from the default view. Vizzy now determines the right workflow from context:

What You Do	What Vizzy Picks
Type a description, no images	Text-to-image generation
Upload images + ask for changes	Design-preserving rendering (image-to-image)
Say "animate," "video," or set a duration	Video generation (image-to-video)
Upload 3+ images + uniform change (paid plans)	Smart Batch Generation

If Vizzy is ever uncertain - say you upload several images and it's not clear which are references and which are targets - it asks. No guessing, no wrong assumptions.

If you want to set default parameters - like a preferred resolution or aspect ratio - you can do that in the Settings panel. But you'll never need to pick a workflow manually again.

Chat-Driven Settings

You no longer need to hunt for the aspect ratio dropdown or figure out where the video duration slider lives. Just say it:

Vizzy updates the aspect ratio to 16:9 and confirms with a small inline badge in the chat. The change persists for your next generations.

Vizzy switches the output to video, sets the duration to 5 seconds, and generates - all from one sentence. You see a compact confirmation in the chat showing exactly what changed, so nothing happens behind your back.

This works for any parameter: model selection, resolution, aspect ratio, video duration, audio, loop settings. If it's a setting, you can change it by asking.

Smarter Conversation Memory

Long design sessions are common. You start with architecture, pivot to interiors, explore a completely different concept, then circle back. The old system kept a fixed window of recent messages - which meant Vizzy could lose context from important earlier decisions.

The new system is smarter about this. It keeps your most recent messages intact and summarizes older parts of the conversation into structured context. So when you say "go back to the villa concept from earlier," Vizzy knows what you're talking about.

It also tracks when you change reference images mid-conversation. If you swap out a sketch at message 12, Vizzy knows the old sketch is gone and won't accidentally reference it.

Faster, More Reliable Generations

This one's under the hood, but you'll feel it. Previously, generation involved a round-trip: the AI decided what to generate, sent that decision back to your browser, your browser called the server to start the actual generation, and then fed the result back into the conversation.

Now it's direct. Vizzy decides what to generate and starts the generation server-side in the same step. One trip instead of three. The result streams back to you as it happens - you see skeleton cards that fill in with your images as they complete.

This means fewer stuck loading states, fewer failed generations from network hiccups, and render usage that's counted more securely because it happens on the server, not in your browser.

A Cleaner Interface

The chat form has been stripped to essentials:

• Message input - where you describe your vision
• Image upload - drag and drop your references, always visible
• Model selector - compact, one click
• Send button

That's it. No workflow cards, no inline parameter controls cluttering the conversation. If you want to set default generation parameters - resolution, aspect ratio, video duration, and similar - there's a dedicated Settings panel where you configure those once. Vizzy uses your defaults as a starting point and overrides them intelligently based on the conversation.

Batch Generation, Now Fully Conversational

On every paid plan, Smart Batch Generation is now fully integrated into the conversational flow. You don't need to switch batching on - Vizzy detects it from your images and prompt.

Upload 8 product photos and say "change all backgrounds to a minimalist white studio." Vizzy sees 8 images with a uniform transformation request and automatically runs a batch - one generation per image, applied individually.

Upload 6 architectural renders plus one mood board and say "apply the mood from the last image to all the others." Vizzy identifies the reference, applies it to each target, and runs the batch.

The intelligence that used to require you to understand batch patterns now lives inside the agent. You just describe what you want.

What Stays the Same

Not everything changed. The core systems you rely on are untouched:

• Generation quality - same rendering pipeline, same AI models, same output quality
• Your projects and history - all existing chats and generations are preserved
• Pricing - same costs, same calculations, now counted more securely server-side
• Webhooks and real-time updates - images still appear via real-time subscription as they complete
• Old chats - previous conversations display correctly with full backward compatibility

If you open an old chat, you'll see your history exactly as it was. New messages in that chat use the new agentic system automatically.

What This Means for Your Workflow

The practical effect is fewer decisions before you start creating. Instead of choosing a workflow, setting parameters, uploading images, and then describing what you want - you describe what you want. Vizzy handles the rest.

For quick concept exploration, this means faster iteration. Describe a space, get a render, ask for a video of it, change the lighting to golden hour - all in the same conversational flow, without switching workflows or panels.

For client presentations, this means less technical overhead. Focus on the design conversation, not on configuring a tool. Your chat history becomes a natural record of the design evolution.

For batch workflows, this means no manual setup. Upload your images, describe the transformation, and let the agent figure out whether it's a multi-image batch, a reference-based transfer, or a multi-view exploration.

Try It Now

The update is live. Open any project and start a conversation - Vizzy will take it from there.

You can set your preferred defaults in the Settings panel, but you won't need to pick a workflow or tweak parameters before every generation anymore.

Start a Conversation with Vizzy

---

Frequently Asked Questions

Do I need to do anything to get the update? No. The new agentic system is live for all users. Open your project and start chatting.

Can I still set my preferred parameters? Yes. The Settings panel lets you configure default parameters for image generation (resolution, aspect ratio) and video generation (duration, audio, loop). Vizzy uses these as defaults and overrides them when the conversation calls for it.

Will my old chats still work? Yes. Old conversations display correctly. When you send a new message in an old chat, it uses the new agentic system. No data migration, no broken history.

Is Smart Batch Generation still Max tier only? No. Batch generation is on every paid plan: up to 4 images at once on Hobby, 8 on Pro and 16 on Max.

What if Vizzy misreads my intent? It asks when it's unsure. If it does get it wrong, just tell it in the chat: "Render this faithfully to my upload." Vizzy adjusts immediately. You can also change your default parameters in the Settings panel to guide its choices.

Did render costs change? No. Generation costs are identical. The only change is that usage is now counted server-side instead of client-side, which is more secure and reliable.