Use cases

How far it actually goes, on four real cases. Proofs of reach, not tutorials for someone else's tool.

These pages answer one question: what can this actually do that a text box cannot? They are proofs of reach, written from work that was really done, and they name no brand — the tool you point Diwall at is yours, not ours.

What the web actually answers

Before the cases, a number worth having. On 27 June 2026, across a campaign of 23 real commercial pages, fewer than one page in ten answered. That is the honest backdrop to every case in this section, and the reason none of them promises that anything will simply work. The full record, with the follow-up benchmark, is in Web accessibility.

What they have in common

Each case is a task somebody needed done — administering an interface, diagnosing a local page, reading and extracting, dealing with a site that refuses. None is a demonstration built for the occasion, and each one names what it cost, including the attempts that failed.

In detail

Diagnosing an interface you are building — A layout that shifted, an interaction that stopped working. The only case here that ships as a runnable scenario — you can execute it before reading further.

Reading pages and extracting what is on them — Comparing an item across several shops, mapping a documentation site, checking a calendar. Read-only work, no vision model, a handful of requests.

Administering a web interface end to end — Configuring a dashboard, running a ticketing platform through its own admin panel. Multi-step, authenticated work — including the frictions that showed up along the way.

What the web refuses to an agent that says what it is — An agent that says what it is gets turned away by a large share of commercial sites. Not a defect to fix — it is the finding.