Generic image models treat a facade sketch, a car three-quarter view, and a staged living room as the same problem. You end up fixing prompts, fighting materials, and re-explaining scale until the client meeting.
Visualizee Skills changes that. Every time you message Vizzy, it loads a specialist pack - architect, automotive, fashion, six more - before it generates anything. Same plain-language prompt. Materials that read right, lighting that holds, fewer "why does this car look like a toy" moments.
Same prompt. Specialist on the other side.
What runs every time you hit send
A skill is not a loose tip or a one-off template. It is structured instruction Vizzy injects into the session: how to talk about materials, which lighting setups to prefer, what to preserve from your upload, and what to avoid (warped mullions, floating furniture, chrome that reads as plastic).
Every turn starts the same way:
Shared baseline - studio fundamentals (photorealism, composition, iteration language) plus motion craft when you are in video mode.
One persona pack - nine options today, chosen to match your project and ingredients.
Visible confirmation - a compact strip in chat listing the active persona, intent, and every loaded skill.
No hidden reasoning block. No digging through settings to see what Vizzy is using.
Nine specialists. No generalist.
There's no "generalist" catch-all - on purpose. Each persona maps to how design teams actually work.
Persona
Built for
Skills in the pack
Architect
Facades, massing, site context
Building exteriors, Site & urban context
Interior designer
Rooms, FFE, interior light
Interior spaces, Materials & FFE
Automotive designer
Vehicle design, studio, CMF
Vehicle exterior design, Studio lighting & CMF
Tuning & aftermarket
Mods, stance, liveries
Tuning & aero, Wheels & livery
EV & concept
Future vehicles, design language
EV concept vehicles, Design language
Industrial designer
Products, appliances, packshots
Product hero shots, Industrial & appliances
Fashion photographer
Editorial, catalog, portrait
Fashion editorial, Lookbook & catalog, Portrait & people
Real estate
Staging, listing photography
Interior staging, Exterior & twilight
Landscape architect
Gardens, planting, outdoor design
Landscape & gardens
Shared skills load on top of whichever persona wins: Studio fundamentals for image work, Motion & video when you are generating movement.
That's 9 personas, 21 specialist skills, and 2 shared layers. Every pack is documented inside the app, so you can read what Vizzy will use before a client call.
Picked for you, every message
Visualizee Skills sit on top of the agentic Vizzy workflow. You do not pick a pack from a dropdown before typing.
At the start of each turn, Vizzy runs discoverSkills: it reads your message, your uploaded ingredients, and the session context, then commits to one persona and an intent (create, iterate, animate, inspire, or explore).
You say / upload
Typical pack
Facade sketch + "golden hour villa"
Architect
Empty room photo + "stage for listing"
Real estate
Stock coupe + "widebody, lowered, matte gunmetal"
Tuning & aftermarket
Clay model + "studio sweep, paint reflection"
Automotive designer
Last render + "5-second dolly, keep materials"
Prior persona + Motion craft
If the job spans disciplines - say a mixed-use building with a hero interior shot - Vizzy picks the persona that best matches the primary deliverable for that message. The next message can shift packs when your focus shifts.
What changes for you
Stop pasting "photorealistic, 8k" on every message
Studio fundamentals already set the quality bar. Persona skills add the domain vocabulary - glazing behavior, FFE scale, tire stretch, listing-camera height. The prompt boilerplate you've been copying for two years stops earning its keep.
No more plastic wood, floating sofas, or fake chrome
Architect packs bias toward believable facade assemblies. Interior packs weight fabric, stone, and wood grain. Tuning packs understand livery continuity across panels. The specialist failure modes get filtered out before they render.
Show clients what the AI is actually using
The skill strip answers a question clients rarely ask out loud: what's the AI doing here? Point at the active pack in the thread. Useful in regulated workflows where traceability matters.
A studio playbook your juniors can read
Open Skills in the app sidebar to browse every pack and read the full instructions. Use it to onboard new hires or align a studio on what "good output" means for each job type.
Skills + Agentic Vizzy + Batch Work
Visualizee Skills define how Vizzy thinks about craft. Agentic mode defines what workflow runs (Inspiration, Render, Motion, batch). They stack.
Example on Max tier: upload six product angles, ask for the same studio sweep on all of them. Vizzy selects Industrial designer craft, detects batch intent, and runs Smart Batch Generation - one pack, one workflow, six consistent outputs.
Example for architects: upload a plan, ask for a furnished 2.5D plan view. Architect pack keeps plan geometry; Render mode handles the transformation. You never typed "top-down, preserve walls."
Who Benefits Most
Architects - facade and site packs for competitions and client reviews.
Do I pick the persona manually?
No. Vizzy selects it via discoverSkills each turn. You can steer with clear language ("treat this as a listing photo," "this is a widebody build") if the wrong pack appears.
Can I pin a persona for every message?
Not yet. Selection is automatic per turn so packs stay aligned with what you are asking right now.
Where do I read the full skill instructions?
In the app: Skills in the sidebar, then open any skill for the complete body text.
Do Visualizee Skills cost extra credits?
No. They are part of the Vizzy session - the same generation pricing as before.
How is this different from Prompt Assistant?
Prompt Assistant helps you write a single message. Visualizee Skills are system-level craft loaded for the whole turn, before generation runs.
Will more personas be added?
We add packs when a discipline needs a stable, testable instruction set - not as one-off templates. Watch the changelog for updates.
