Skip to content

Feature · Relay Assistant

AI that helps you make the screen.

Generate artwork, read existing content, and propose layouts, copy, or structured changes inside the same workflow where you edit and publish. The output stays reviewable instead of jumping straight onto a TV.

Real review workflow

AI proposes. You inspect the structured result.

The product keeps imported rows, confidence feedback, edits, and publishing decisions inside the operator workflow—not behind an automatic publish button.
  • Photo or PDF becomes reviewable structured content
  • Accept or correct values before records are created
  • Nothing reaches a live screen until you publish

REAL PRODUCT INTERFACE

Canvas Relay data panel showing editable imported menu records before publishing
Real review workflowCaptured from Canvas Relay
Cloud basedAccess anywhere
Design onceUpdate in real time
Relay everywhereAny screen, any size
Reliable & secureOffline playback built in

What Relay Assistant does now

01

Generate signage-ready artwork

Prompt for a landscape, square, or portrait image and save the result into Media. Use it as a background, visual, product atmosphere, event backdrop, or ambient art after you review it.

02

Photo/PDF → editable content

Vision AI reads printed menu content into structured sections, items, and prices with a review step before records are created. Pasted-text parsing remains a model-free fallback.

03

Editable scene candidates

A general prompt creates two or three reviewable scene candidates inside your existing board using supported text, shapes, and explicitly authorized Media images. Accepting a candidate is explicit and atomic.

04

Copy suggestions

Request description/copy help where the current structured-content workflow supports it, then accept only the suggestion you want to keep.

05

Natural-language price intent

Instructions such as raising a set of prices can be interpreted into a server-computed from→to diff that stays visible before it is applied.

06

Grounded product help

Relay Assistant can answer product-support questions from Canvas Relay documentation and use explicitly shared diagnostics rather than inventing screen state.

Creator

30

AI credits / month

Pro

100

AI credits / month

Studio

500

AI credits / month

Current boundary: general prompts create two or three editable scene candidates inside an existing board. They do not yet generate an entire multi-scene board with schedules, widgets, menus, and data integrations. Every candidate remains reviewable, acceptance is explicit and atomic, and AI never publishes.

Questions, answered

01Can Canvas Relay generate images for a screen?+

Yes. Relay Assistant can generate signage-ready images in landscape, square, or portrait formats and save the result to your Media library. It does not place the image on a board or publish it automatically — you decide where it goes.

02How does photo/PDF menu import work?+

Upload a photo or PDF. AI extracts sections, items, and prices with confidence information, then you review and edit the rows before accepting them as linked records. Pasted text can be parsed without model credits.

03How do AI credits work?+

Creator includes 30 AI credits per month, Pro includes 100, and Studio includes 500. Model-backed actions reserve credits before they run; deterministic tools such as pasted-text parsing and direct structured edits do not consume AI credits.

04What happens when I use all my AI credits?+

The editor, boards, menus, publishing, and screens keep working. Model-backed AI actions pause until the allowance resets or the plan changes; manual and deterministic editing remain available.

05Can AI change what's on my TVs without me knowing?+

No. Generated media is saved separately, and layout/copy/data actions are proposals you review and accept or discard. Canvas Relay does not let the model silently publish to a live screen.

06Can I describe a screen and get editable scene ideas?+

Yes. A general prompt creates two or three reviewable, editable scene candidates inside an existing board. Supported general candidates use text, shapes, and Media images you explicitly authorized. Acceptance is explicit and atomic, and AI never publishes.

07Does one prompt create a complete multi-scene board?+

Not yet. One prompt does not currently assemble an entire board with multiple scenes, schedules, widgets, menus, and data integrations. Canvas Relay describes the implemented editable scene-candidate workflow without calling it complete board generation.

Use AI for one part of the screen — not the final say.

Create a free account, open Relay Assistant, and keep every generated or proposed change under your control.

Try the editor — no account needed

Free plan includes one live screen. No credit card. No trial clock.