Where AI test authoring should land first

Everyone is building AI that writes web tests; the crowded direction is probably the wrong first move. This note holds the thesis that mobile app testing — where authoring expertise is scarcest and tooling competition thinnest — is where AI test authoring earns its keep first.

Problem

Natural-language test creation is becoming table stakes on the web, with aggressive open tooling capturing that ground. Meanwhile mobile automation adoption stays low precisely because authoring expertise is the barrier — the teams that most need generated tests are the least equipped to write them today.

Where it leads

This is research work — it ends in open threads, not outcomes. They're below, stated plainly.

The reasoning is a positioning argument. Web-side authoring is contested by fast-moving open tooling, and the platform’s web usage concentrates on one browser engine anyway — differentiation there is a knife fight. Mobile app automation is the inverse: adoption sits at withheld: adoption share , a large cohort of teams runs withheld: dormant-cohort share mobile sessions at all, and the named barriers in the research are development knowledge gaps and integration friction — exactly the deficits generated authoring compensates for.

The design principle the thesis commits to: AI authoring should land where expertise is the barrier, not where typing is the barrier. On the web, natural language saves keystrokes for people who could have written the test. On mobile, it creates testers who otherwise wouldn’t exist. Same technology, different theory of value — and only one of them changes the adoption curve.

This is the corpus’s youngest note, and it’s deliberately thin: a thesis, a drafted user-story matrix, and a queue of assumptions to kill. The threads below are the honest state of it.

Open threads

The thesis is unproven where it counts: does a mobile-first authoring tool meet teams at their knowledge gap, or does the gap swallow the tool too?
Authoring can't be a demo that decays — generated tests must survive app churn, or AI authoring becomes AI maintenance debt.
The user-story matrix is drafted; the honest next step is watching a dormant-adoption team try to author their first mobile test with nothing but intent.
Where authoring meets the existing healing and failure-analysis agents is an architecture question with UX consequences — one assistant or three tools?

The full walkthrough — screens, numbers, names — happens in conversation. Start one →