Skip to content
Pressline

AI-assisted Engines

An AI Engine differs from a template Engine only in where the pixels come from. Keep the model behind a small interface (the sample’s AiAdapter: generate(prompt) → { bytes, contentType }), store the raw result under the Design, and feed it to the same render() path per Spec. Practical points:

  • Ask the model for the design’s aspect (portrait for tees and posters) and the largest size it offers; upscaling a small image to a 4500×5400 DTG file shows.
  • Set sellable: false until moderation or the Customer’s confirmation has run; Pressline will not sell it.
  • Keep prompts and raw outputs private; only the Preview and Printfiles are public URLs.
  • Generation is slow: answer 202 to the printfile request if you render on demand, or pre-render on finalize.
  • The sample Engine enables an OpenAI adapter when OPENAI_API_KEY is set; without a key it is a plain template designer.