AI Studio (Prototyping)

Google AI Studio is our go-to tool for fast prototyping: turning a design idea or a Figma design into a clickable, working prototype quickly. It now runs on LowCode's own company billing, so use it freely for work.

Why we use it

It is the fastest tool the design team has found for prototyping: build something in AI Studio and you have a working, clickable app in minutes, closer to real software than a static mockup. Enrique's Trade Ready prototype (trade-ready-gamma.vercel.app) is a good reference for how a finished prototype should be organized.

Set up company billing (do this once)

Log into aistudio.google.com with your @lowcode.agency account, not your personal Gmail. If you're already signed in with a personal account, use a separate browser profile or an incognito window.
Go to Dashboard → Projects and import the project lowcode-ai-studio (shows as "LowCode AI Studio"). It will not show up in the project picker until you import it.
Select it in the project dropdown at the top.
If you already have personal API keys in use, click the billing alert, select Gemini API and Continue with pay per request, switch the project to LowCode AI Studio, name the new key (e.g. "Yourname - AI Studio") and click Next. If you run out of credits inside an existing project, you can do the same thing from Settings in the top right.
Any API key created inside that project bills to the company automatically.

Rules

Never enter a personal credit card. You don't need one. If a screen asks for payment info, stop and ping Jesus.
Usage is billed to LOW / CODE, so it's real money. Budget alerts are on. Use it freely for work, just don't leave batch jobs running.
On this project your prompts are not used to train Google's models. That matters for client work.
If you see "you do not have permission to create a key in this project," ping Jesus and he'll fix the access.
If something you built lives in a personal account, move it over to the company project when you get a chance.

Prompting: the RTCCF framework

When prompting AI Studio (or similar tools) to recreate a Figma design as a prototype, structure the prompt with these five parts:

Role: who the AI should act as, e.g. "You are a senior UI designer working for [client], specialising in prototyping and design systems."
Task: what to build, e.g. "Recreate the provided UI matching exactly what we have in Figma, following a structured, component-based approach."
Context: what the screen/feature actually is and what it needs to show.
Constraints: spacing rules, reusable components, accessibility, typography, no hard-coded values where components can be reused.
Format: auto layout, components from frames, clean layer naming, ready for design system integration.

Known limitation

Credits and rate limits on AI Studio are not clearly documented by Google, and a personal Google AI Plus/Pro subscription does not carry over credits to AI Studio or Antigravity (only Gemini app usage). Using the company project above avoids this: it bills per request from the first API call rather than relying on a personal plan's quota. If you hit a wall anyway, switch to the LowCode AI Studio project/key.