Skip to main content

Add AI inside your app with Mythex

This is AI as a feature of the app you ship (a support chatbot, summarizer, generator). It is separate from the Mythex coding agent.
Provider usage bills your OpenAI / Anthropic / other account. Store keys in project secrets. Never expose them to the browser.

When this is a good fit

  • The product needs chat, classify, or generate-from-text
  • You already have (or can create) a provider API key
  • You can keep the first version to one AI action (e.g. “summarize this ticket”)
Skip this if you only needed Mythex to build a normal site — you don’t need an LLM in the published app.

What you’ll have

  • A UI (chat panel, “summarize” button, etc.)
  • A server-side call using a secret env var
  • Optional: save threads in a database

Starter prompt

More detail: Add AI features with your own API keys.

Build it

  1. Create the key at your provider and add it in project secrets (or via secrets in chat).
  2. Prompt for one feature. RAG, tools, and multi-model routers can wait.
  3. Confirm in Preview that the key never appears in client-side code.
  4. /test send a message → a response appears (you’ll need the secret set).
  5. Publish — published apps need the same secrets available at runtime.

Honest limits

  • Mythex does not include a built-in “app AI” SKU with included tokens for your users
  • No native knowledge-base product — you bring data and the API
  • Rate limits, safety, and cost controls are yours to design