Product details

Prompt Canvas detailed guide

Prompt Canvas creates implementation instructions from screenshots. It does not modify the source code of a website or application. It organises visible intent and hands it to the AI, developer, or development environment responsible for implementation.

From screenshot import to implementation handoff

Step 1
Import the screen to be changed
Load a PNG, JPEG, or WebP from a file, drag and drop, or the clipboard. Choose a use case and restore saved work when an earlier canvas should be continued.
Step 2
Show the target and intended result
Move, resize, duplicate, or delete a selected element, or add text, arrows, shapes, highlights, blur, and other marks to express the change directly on the screen.
Step 3
Record intent and acceptance criteria
Capture not only where the change happens, but also the before and after state, priority, reason, and acceptance criteria. The canvas updates the prompt and structured change data.
Step 4
Review and hand off
Choose the prompt, change JSON, annotated image, source image, report, or ZIP required by the next tool. When the optional experimental AI-analysis feature is used, review the destination and payload before sending.

Practical use cases

Communicate a layout change

Select a card or button, show its intended destination and size on the screen, then hand the change to the implementation owner.

Combine review comments into one handoff

Assign priorities and acceptance criteria to several visible issues, then share the annotated screen and change list together.

Hide personal information before asking for help

Blur or mosaic names, email addresses, and other sensitive areas, then review the content included in the export.

What stays local and what happens in another tool

Available without an AI connection

Instruction authoring and export run in the desktop application.

Image import and visual editing
Project save, autosave, and version comparison
Prompt and change JSON generation
Annotated image, report, and handoff ZIP export

Environment or destination required

Analysis and implementation are completed by supporting tools.

OCR in a supported Windows WebView environment
A cloud API key or Ollama connection for experimental optional AI analysis
An AI, developer, or development environment that receives the handoff
Visual and functional verification after implementation

Capabilities and requirements

What changes with this product

Show a change where words are not enough

Place arrows, text, shapes, highlights, movement, or privacy treatments directly on a screenshot. Store position, size, intent, priority, and acceptance criteria with the visual instruction.

Convert visual feedback into an AI-ready handoff

Canvas changes continuously update a prompt and structured change JSON. Package them with the source and annotated images for ChatGPT, Claude, Gemini, Codex, or another supported destination.

Keep the image and instructions under local control

Editing, projects, autosaves, version comparison, and exports stay on the PC. An image is sent only when the user explicitly runs the experimental AI-analysis feature after reviewing the destination and payload.

The full capability set

Import and project management

Bring a screen into the workspace quickly and preserve work in progress.

PNG, JPEG, or WebP from file, drag and drop, or clipboard.
Undo, Redo, autosave, and backup recovery.
Named saved versions with notes and comparison.
Layer order, visibility, locking, and multi-selection.

Visual editing tools

Add instructions in forms that remain readable on top of the screen.

Drawing, erasing, text, arrows, circles, rectangles, and highlights.
Blur, mosaic, crop, and whole-image resize.
Move, resize, align, distribute, and group elements.
Grid, snapping, zoom, brush opacity, and smoothing.

Structured instructions

Translate visual changes into information an implementation workflow can use.

Pixel and normalised coordinates.
Before, after, priority, intent, and acceptance criteria.
OCR candidates in supported environments.
Provider-specific prompt formatting.

Handoff to AI or a developer

Inspect the content and export the format that suits the next tool.

Prompt, change JSON, annotated image, and source image.
AI handoff ZIP with a manifest and image SHA-256 hashes.
Shareable HTML report with the image and change list.
Instructions for ChatGPT, Claude, Gemini, Codex, OpenRouter, and Ollama.

Requirements and support

Operating system: 64-bit Windows 11.
Processor: 1 GHz or faster, two or more cores, x64.
Memory: 4 GB minimum. 8 GB or more is recommended for large images and multi-layer editing.
Free storage: At least 500 MB for installation, plus additional space for images, projects, autosaves, and exports.
Display: At least 1100×720. 1280×800 or larger is recommended.
Additional component: Microsoft Edge WebView2. Its setup process is included with the installer.
Internet: Not required for local editing. Required for cloud-AI analysis, API-key acquisition, update checks, enquiries, and purchase.
Image input: PNG, JPEG, or WebP up to 25 MB through file selection, drag and drop, or clipboard paste.
Project format: .promptcanvas project files up to 40 MB, validated for schema version and content when opened.
Exports: Prompt, change JSON, annotated image, source image, shareable HTML report, and AI handoff ZIP.
AI analysis: Optional and experimental. It is not required for the supported prompt-generation and export workflow. It requires a configured cloud API or local Ollama connection and presents the destination and payload before sending. Provider APIs, models, and fees may change.

Frequently asked questions

Does Prompt Canvas change the website or application itself?

No. It creates implementation instructions. Source changes are carried out by the AI, developer, or development environment that receives the handoff.

Can I use it without connecting AI?

Yes. Visual editing, prompt and JSON generation, projects, annotated images, reports, and ZIP exports work locally. Only the experimental AI-analysis feature requires a provider connection.

Are images uploaded automatically?

No. Connection checks do not upload an image. An image is sent only after the experimental AI-analysis feature is selected and the outbound request is reviewed.

Is OCR always available?

OCR depends on TextDetector support in the Windows WebView environment. Manual text layers remain available when OCR is unavailable or inaccurate.

Which AI tools can receive the handoff?

Prompt formats are provided for ChatGPT, Claude, Gemini, Codex, OpenRouter, and Ollama. The destination still needs suitable access and capabilities.

Price and included items

Prompt Canvas for Windows

US$29.99 · one-time purchase

Japan: JPY 4,980 including tax · one-time purchase

For one individual or business user

One user may install the app on up to two Windows PCs under their control. Images, instructions, and reports created with Prompt Canvas may be used commercially. The application itself may not be redistributed or resold.

Prompt Canvas 1.0.0 release for 64-bit Windows 11
Minimum: 1 GHz two-core x64 CPU, 4 GB RAM, and 500 MB free storage
1.x updates and a basic setup guide
Support through the enquiry form, normally answered within three business days
Local editing plus prompt, JSON, image, ZIP, and HTML exports without an AI connection
Direct AI analysis is an optional experimental feature; third-party API charges are not included
Installer SHA-256: E1D745ABC1F5F2819817A49DCCCE2E1EDE8330F535FC6A2108D4295AB989F472

Purchase for US$29.99 — SAINO official Buy on Gumroad Ask a question before purchase

From purchase to installation

01
Review the purchase conditions
Confirm the price, device allowance, unsigned status, system requirements, delivery timing, and cancellation and refund terms.
02
Choose where to pay
Choose SAINO checkout powered by Stripe or Gumroad, then confirm the product, currency, total, email address, and billing details before paying in full. Card information is never entered inside Prompt Canvas.
03
Receive the product
SAINO checkout emails a time-limited download within three business days. Gumroad provides the installer, SHA-256 checksum, and setup guide from the purchase page or receipt email.

Minimum requirements: 64-bit Windows 11, a 1 GHz two-core x64 CPU, 4 GB RAM, 500 MB free storage, and a 1100×720 display area. 8 GB RAM and 1280×800 or larger are recommended. Internet access is required for experimental cloud-AI analysis, API-key acquisition, update checks, enquiries, and purchase. A full refund is available before the download URL is sent. Customer-choice refunds are not available after delivery; report a SAINO-caused blocking defect within 14 days.

Let’s turn an idea into something real.

Ask about a product or a custom website, web application, desktop application, or software project. Early ideas are welcome.

Discuss a project