Every ai photo editor on the market offers to fix the picture for you, and this one does the opposite: it never touches your file. It is a coach. You send a screenshot of Lightroom, Photoshop, Canva or Snapseed with the sliders visible, and it reads what is on screen and hands back the recipe: which panel, which control, which direction, how far. You move the slider, so the photo stays yours and nothing is baked in.
A session runs as one loop. You show the current state, it names exactly one action by its on-screen label, you do it, you show the result. Never two steps at once, because the second depends on how the first one lands. What you are buying is the interface map inside the file: real panel names and menu paths across Lightroom Classic and cloud, Photoshop, Canva and Snapseed, the five beginner mistakes and how each looks in a screenshot, and a twelve step arc from white balance to export.
The same file works in Claude, ChatGPT and Gemini. Paste it once into a Project, a custom GPT or a Gem, which takes about two minutes, and it is a single payment with nothing to renew.