Skip to main content

AI usage in ScreenCloud Arc

An overview of where AI features live in ScreenCloud Arc, with links to the full details for each one.

Arc uses AI in two areas: writing posts, and processing feed content. This article outlines each, with a link to the full details.


AI in Post Writing

The AI Writing Assistant sits inside the post editor and works in two directions: generating a post from a short prompt, or editing text that's already there.

Starting from Nothing

A set of ready-made message types covers the posts sent most often, including Announcement, Praise a colleague, and Birthday celebration, but the Assistant isn't limited to these. It can help draft any message: describe what you want to say and it'll write the post.

Editing Existing Text

For text already written or pasted in, the Assistant can adjust tone, translate it into another language, or summarize it down to what's worth showing on a screen. Requests aren't limited to these three: it also handles specific asks such as shortening a paragraph, reworking a headline, or adding a call to action.

Summarizing Links and Documents

The Assistant can also summarise content that's attached rather than typed: the page behind a shared link, or an attached PDF.

Document limit: PDF summarisation works for documents of up to 10 pages. Longer documents can still be posted and shown as a QR code, but can't be summarised.

See Creating Posts for Your Screens for the full details.


AI in Feed Processing

When a feed is connected, an AI Instruction can be applied to shape its content before any of it reaches a screen. Instructions can be written from scratch or picked from a built-in template library, organised into three types:

  • Rewrite: changes how a post reads, for example summarising it for signage, simplifying the language, or shifting the tone.

  • Filters: controls what gets shown at all, for example surfacing announcements only, or filtering out noise and bot messages.

  • Translations: translates incoming posts into a specific language.

Instructions can be combined, for example "Translate from English to Thai and remove any posts about animals."

Testing an Instruction

Before saving, an instruction can be checked with Test Run, which applies it to existing content without saving anything. Results show in two views: a list showing how many posts would be shown versus hidden, and a side-by-side comparison of the original post against the AI-applied version.

Timing: AI Instructions only apply at the point a feed pulls content in. Changing an instruction afterwards doesn't affect posts already collected. Instructions are text-only, so they can't be used to remove an image while keeping the rest of the post.

See Connecting Feeds to Your Screens for the full details.


AI in Translation

Translation is one function of an AI Instruction applied to a feed. It performs an active translation of the text, rather than displaying whatever language the source happens to use, which is the default behaviour for feeds and apps without an instruction applied.

A translated post shows a note on which language it was translated from, with a Show original toggle to check the source text. Translation instructions can be combined with a filter or rewrite instruction in the same feed.

See Which Apps and Feeds Support Multi-Language Content? for language-specific detail, and Connecting Feeds to Your Screens (linked above) for how to set up a translation instruction.

Did this answer your question?