Arwell

Arwell comes out of AAAnow, the London group that has been automating digital quality since 1999, and whose AiSC Scorecard and Arty I had worked on before. The product does one thing. It finds every PDF linked from an organisation’s websites and publishes an accurate web version beside each one, so a person on a phone, someone using a screen reader and an AI assistant answering on their behalf all read what was actually said. The original file is never touched.

What you publish should be what AI repeats.

Paper, filed

The site has to sell a compliance product to councils, universities, pharmaceutical companies and publishers without sounding like one. So the palette is paper: a warm off-white and sand, deep green for the dark sections, and one red-orange reserved for the words in a headline that carry the point. Headlines are set in Faculty Glyphic, with its flared, cut-in letterforms, body copy in Geist, and every label in IBM Plex Mono. Each platform page carries a reference and a date in that mono — ARW/PLT/2026/14, updated 9 October — like a document in a register, which is exactly what the product keeps.

Artwork made of documents

Instead of product screenshots, the imagery is drawn live on canvas: collages of scanned text, old engravings and colour grids, a scatter of document reads converging on a line, red seven-segment clocks. The hero runs a crawl log in its margins — PDFs found, web versions published, images described, tables rebuilt — so the first screen shows the work being done rather than describing it. The robot who greets visitors on the approach page is Arty, AAAnow’s mascot, kept to a cameo here.

Buying in the open

Arwell does not do sales calls, and the site is built around that. Pricing is public: dials for monthly page views, sector, service level, term and payment type shape a proposal on the page itself. A five-step business-case builder asks what problem you are solving, how far you want to go and who the proposal is for, and a written proposal follows within a day, to be forwarded internally and accepted online. The whole journey is written down on a page of its own, six stages, every one of them the buyer’s to stop.

A page for every reader

Beyond the landing page there are six use cases, six platform pages — the secure tag, analytics, editing, the AI view, three views of a document and the infrastructure schematic — a page on what governments in the UK, US and EU recommend, a partner programme, a help library, a blog of notes on documents and who reads them, and a sign-in by magic link or passkey. The front end is built in Vue.