AI Transparency

Ideairy is built on top of multiple frontier AI models. We believe you should know exactly what powers the websites we generate for you. This page lists the providers, what we use them for, and how your data flows through them.

Providers we use

• Anthropic Claude

  claude-sonnet-4.5, claude-opus-4

  Reasoning, planning, and high-quality copywriting for landing pages.

• Google Gemini

  gemini-2.5-pro, gemini-2.5-flash

  Primary site generation, multi-step coding, and large-context edits.

• OpenAI GPT & Codex

  gpt-5, gpt-5-mini, gpt-5-codex

  Code generation, refactors, and structured output for complex components.

• Nano Banana (Gemini Image)

  gemini-2.5-flash-image

  On-demand image generation and editing inside generated sites.

• Supabase

  Postgres 15, GoTrue auth, Storage

  Authentication, Postgres database, storage, and serverless functions.

How your prompts are handled

When you send a prompt, it is routed to the AI model best suited for the task. Prompts and generated files are stored in our database so you can revisit your projects. We do not sell your data, and providers process requests under their published enterprise data policies (no training on your inputs).

Model selection

We default to Google Gemini 2.5 Pro for full project generation because of its large context window. Smaller edits, chat replies, and image generation may use Gemini Flash, GPT-5, Claude, or Nano Banana depending on the task. The active model is recorded with every message for full traceability.

Your responsibility

AI output is provided "as is". Always review generated code, copy, and images before publishing — especially for legal, medical, or financial content. See our Terms of Service for the full agreement.

Contact

Questions about how we use AI? Email help@ideairy.ai.