Why I rebuilt HerdPrint’s Mac workspace for the web
Most app pages ask for a download before the visitor has learned how the product feels. That is a large commitment for someone who only wants to test a layout or click through a workflow. I built a browser version of HerdPrint’s main workspace so that decision can happen first. The interactive demo opens immediately, responds to real choices, and asks for the Mac app only when the visitor reaches a native task.
This is a build story about the gap between seeing software and trying it. HerdPrint happens to print PDFs and images, but the problem applies to many desktop utilities whose value lives in the relationship between several controls. One headline feature cannot explain that relationship.
A screenshot flattened the part I wanted people to understand
HerdPrint has a three-part workspace. Files and date groups live on the left. The current sheet sits in the center. Layout, orientation, scale, and output controls live in the inspector on the right. A change in one area alters what the other two show.
That feedback loop is the product. A gallery image can show the three columns, and a recorded video can show my preferred path through them. Neither lets a visitor switch from two stacked documents to a nine-item grid, turn the paper sideways, alter scale, or open the totals panel in an order I did not choose for them.
I wanted the website to answer a practical question: “Would I be comfortable arranging a real batch here?” A static product page could not answer it.
What 1:1 means in this demo
I use “1:1 reconstruction” to describe fidelity to the current Mac workspace: its geometry, color tokens, labels, visual hierarchy, and the way its main controls respond. It does not mean that a browser tab has become a native macOS application, and it is not a claim that every rendered pixel passes an automated comparison.
| Surface | What a visitor can evaluate | What remains unavailable |
|---|---|---|
| Product screenshots | Window structure and visual style | Continuous interaction and personal choices |
| Recorded walkthrough | A complete path chosen by the developer | Exploration outside that path |
| HerdPrint web demo | Layout changes, grouping, totals, watermark settings, print masks, and custom regions | Real files, on-device recognition, system printing, and saved output |
| HerdPrint for macOS | The full workflow from local import to print or export | Requires an App Store download and installation |
A simplified marketing widget would have been easier to build. It would also have introduced a second interface that HerdPrint users never see again after downloading. Matching the real workspace lets the same spatial model carry across the handoff.
DeepSeek helped with the reconstruction, then the prototype had to become a site feature
I used DeepSeek to recreate the existing SwiftUI workspace in HTML, CSS, and JavaScript. The instruction was deliberately narrow: follow HerdPrint’s interface instead of redesigning it for the web. The first useful version lived in one large HTML prototype. Keeping everything together made visual comparison fast while the proportions, rails, inspector rows, and paper canvas were still moving.
Once the behavior settled, I separated the prototype into an Astro route, a stylesheet, an interaction layer, and a six-language dictionary. That change was less glamorous than the visual work, but it turned a disposable mockup into something the product page could link to and the static site could build, index, and maintain.
The browser version now runs on a real state model. A sequence of screenshots could not produce the same feedback. Ten fixed sample documents expose the differences between layouts and date groups. The layout engine computes pages, slots, and fitted document frames. Amount reading advances item by item and leaves uncertain values pending. Language changes rebuild the open summary, amount entry, or export panel along with the main window.
That behavior is why the demo feels closer to the app than a clickable Figma frame. Controls change shared state, and the rest of the workspace reacts.
The stopping point is part of the design
The demo can switch among single, stacked, side-by-side, three-column, adaptive, and nine-per-sheet layouts. It can change orientation and scale, group sample files by date, simulate amount review and totals, edit a tiled watermark, place print masks, and divide a sheet into custom regions. The interface can also switch among Simplified Chinese, English, German, Japanese, Korean, and Traditional Chinese.
It never imports a visitor’s files. OCR, date reading, amount reading, and sensitive-text detection use predetermined sample states. Print, image export, and PDF export stop at a clear download panel. Mail access, printer access, destination folders, and Pro purchasing are absent.
Those omissions keep the promise honest. The web demo answers whether the workspace makes sense. The native HerdPrint app handles local documents and macOS services. The HerdPrint guide remains the source for the complete workflow and current product behavior.
Put the download after understanding
Desktop software often treats the App Store button as the beginning of the explanation. I now think it belongs later for HerdPrint. A visitor should be able to learn the layout model, test the controls, and decide whether the workflow fits before installing anything.
The download prompt appears when someone tries to print or export, which is the moment the browser has reached its honest limit. By then, the request has context. The visitor has already seen how files move through the workspace and what the Mac app would complete.
I previously wrote about how reimbursement printing turned into HerdPrint. That story explains where the features came from. This demo addresses the next problem for an independent developer: making those features available enough to understand before asking anyone to bring another app onto their Mac.
