Interactive photo editing coach built into the browser

The moment this becomes a real problem: a new hire at a small creative agency sits down with a PSD file and a vague brief. They know what the output should look like — they've seen the reference image — but they don't know the path to get there inside the software. Their options are to ask a senior colleague (who is busy), watch a tutorial (which takes longer than just doing it), or guess (which produces bad work). None of these scale.

The reason nothing has properly solved this is structural: it requires knowing both what the user is trying to achieve and what software they're in. Most learning products are either generic (cover all tools in a course) or reactive (answer questions after the fact in a forum). A tool that bridges 'I have this image' with 'here are the specific steps in the tool you're using to get from A to B' requires a combination of image understanding, software-specific knowledge, and step-by-step generation — which was technically awkward until recently.

Adobe has Firefly and in-app AI features, but these generate or modify content — they don't explain how a human would accomplish something manually, which is exactly what users need when they're learning and need to build muscle memory, not just get an output. The complaint 'I needed a couple of tries at the beginning' and 'it took me a few months of almost daily use' reflects a process of trial-and-error that could be dramatically shortened if users could describe what they're trying to do and get a concrete step sequence back — not a marketing explanation of what each tool does.

This is a business because companies with creative teams pay real money for onboarding speed and consistency. A studio that onboards three new editors a year and each takes three months to become proficient is losing billing time. That's a quantifiable cost, and it recurs with every hire, every new project type, and every software update that moves things around.

What to build

Build a web app where a user uploads a before-and-after reference image pair and selects their editing software, and receives a numbered, tool-specific step sequence describing exactly how to achieve that transformation — with keyboard shortcuts and menu paths included for the selected software version.

Where to start

Start with a single, high-volume transformation type — background removal and replacement — because it's the most-Googled photo editing task, has clear before/after inputs, and is performed constantly in e-commerce product photography, giving you a dense enough early user base to collect feedback quickly.

The hard part

Getting the before-and-after interpretation accurate enough to be trusted is hard — if the suggested steps produce a noticeably different result than the reference, users will dismiss the whole product after one try, so the early accuracy bar is both high and difficult to hit without significant fine-tuning.

How it makes money

Per-seat monthly fee for studio teams, with a usage-based free tier (10 queries/month) for individual freelancers who discover it organically and convert as their workload grows.