Every cross-border e-commerce operation I’ve audited carries a hidden tax: process knowledge trapped in one person’s head. The new VA, the new marketplace account manager, the new 3PL coordinator — they all need the same “click here, then here” instructions, but writing those instructions has always taken longer than doing the task itself. StepShot wants to kill that tax. It’s a native Mac app that watches real on-screen work, captures clicks and keystrokes, and produces a polished step-by-step guide as you go. For sellers running remote teams across time zones, the stakes aren’t convenience; they’re accuracy, data privacy, and auditability. If this holds up in practice, it changes how you onboard.
The documentation tax is bigger than it looks
The maker, Thomas Ellon, opens the Product Hunt comments with a line that should feel familiar to anyone who has onboarded a remote employee to Amazon Seller Central or a Shopify store: “I kept writing the same kind of document: click here, then here, screenshot, crop, arrow, number, repeat. An hour of work to describe four minutes of clicking.” Every tool that promised to fix it was a browser extension — useless the moment the workflow touched System Settings, a native app, or a terminal.
That is the exact tax. For cross-border sellers, it isn’t occasional; it’s every onboarding, every policy change, every time the person who “just knows” how to do something leaves. You hand a new contractor access to your store, your supplier portals, and your logistics tools, then spend an afternoon producing a guide that will be out of date by next quarter. Most documentation tools promise to make this easier, but they lean on browser extensions. That fails in e-commerce operations because your team doesn’t live entirely in a browser. Spreadsheets, label printers, FTP clients, PDF editors, VPNs, and marketplace-specific desktop tools all contain steps a VA needs to see before they click.
StepShot’s approach is to capture at the OS level: it watches clicks and keystrokes through the accessibility layer and grabs the screen before each click lands, so the guide shows the button you’re about to press, not the menu that opened after. That is the right mental model for process documentation. You don’t just need a record of what happened; you need the instruction that tells someone what to look for before they act.
The privacy posture also matters. Passwords are never read at all; secure-input fields are excluded at the engine level, so the characters never reach memory. There is no account and no telemetry by default. For anyone who has watched a tool silently upload a screenshot of a customer order, that alone is worth attention.
What StepShot actually does, and the choices I’d defend
On StepShot’s site, the product is described as “AI that turns real workflows into step-by-step guides.” That undersells the mechanics: it captures clicks and typing, local AI frames each screenshot, writes the instruction, and explains which actions to keep or skip. You can edit, annotate, reorder, and redact every step, then export to PDF, Markdown, or HTML. There are two capture modes — action-driven for tight step lists, and full-screen recording when you want the video kept as evidence — plus three capture targets: active window, full screen, or a fixed region.
That feature list alone wouldn’t make me take it seriously. What made me stop scanning was the maker’s “four decisions I’d defend”:
- The preview is the export. The editor canvas, the preview pane, and all three exporters run the same renderer. If you approve one image and ship another, the tool is lying to you. In SOP work, trust depends on what the reader sees matching what you approved.
- Local-first is a philosophy, not a footnote. Projects and screenshots live in a folder on your Mac. The AI is deterministic and local by default. If you want a model to polish the wording, you bring your own OpenAI-compatible endpoint and key, and it only refines a draft that’s already complete. Sending screenshots to a model is a separate, off-by-default consent, and samples are redacted locally before they leave.
- Redaction is destructive. Sensitive regions are resampled down and back up, burned into the bitmap — not a blur or an overlay someone can strip. Local Vision OCR proposes redactions; you confirm each one before export unblocks.
- One-time purchase. Pro is $129, Team is $599 for five seats, perpetual, with all future versions included.
How does this differ from the incumbents on Product Hunt’s own list? Screen Studio produces beautiful screen recordings, not a mutable step list. Guidde makes AI video documentation, which is great for a demo but less useful when a reader needs to pause, redact, and reorder individual actions. StepShot’s bet is that the output should be a document, not a video — and that the document should be assembled while you work, not after.
Why Amazon sellers should care more than the rest
Most of the “document your workflow” attention goes to DTC storefronts, but the pain is much worse on the marketplace side. A seller juggling listing templates, case logs, account health action plans, and reimbursement claims needs exact procedural memory. A step-by-step guide that captures the state of the marketplace’s admin before you click is a compliance artifact, not just a training aid. When an account gets flagged and you need to show a manager exactly how an inbound shipment was created, a screenshot of a button about to be clicked is evidence. The storefront admin world is friendlier, more forgiving, and less likely to send you into a nested maze. So if I had to pick the corner of e-commerce most in need of StepShot, it’s the marketplace operator’s standard operating procedures.
What cross-border sellers can borrow from StepShot’s philosophy
Even if you never install StepShot, the product is a useful provocation. It makes you re-examine how your own operation treats documentation, redaction, and AI consent.
First, portable output. PDF, Markdown, HTML — your SOPs can live in a folder, a Git repo, a wiki, or a Notion workspace. They are not hostage to a proprietary cloud. Cross-border teams should demand that portability from every tool they use, because your VAs are in different countries with different connection speeds, and a locked-in knowledge base is a liability.
Second, local-first AI. StepShot’s default is deterministic and local; cloud AI is an opt-in that uses your own credentials. That is exactly the posture an e-commerce operator should hold with customer data. Too many tools silently send screenshots to a third-party model by default. If a tool can’t show you exactly where data leaves your machine, it isn’t safe to use on a screen that contains customer PII.
The redaction question
The best test for any screen capture tool: what happens to customer data? StepShot’s answer is stronger than most. Passwords never reach memory. Secure-input fields are excluded at the engine level. OCR suggests redactions, but you confirm them before export. And the redaction itself is destructive — it burns into the bitmap, rather than laying a blur over the data. For a seller who has ever copied an order ID or a phone number into a training screenshot, that is the difference between a document and a data leak.
The wider lesson: treat screenshots as sensitively as order records. If a tool’s redaction is reversible, it isn’t redaction.
Where my judgment says it falls short
The honest part is in the maker’s own notes. The maker is honest about what’s not there: no cloud sync, no team accounts, no Notion export, no auto-translation, and Windows is a private test build — not a release. For a solo DTC operator with a MacBook, that’s fine. For a multi-country team where VAs work on Windows laptops, it’s a blocker. You can’t standardize a workflow-documentation system on software everyone can’t run.
The second gap is collaboration. The local-first philosophy is a privacy argument, but it makes life harder when the whole point is distributing guides across a remote team. You’ll end up creating your own sync layer, and your VAs will need their own local projects unless you build a pipeline around the exported Markdown and HTML files. That’s doable, but it’s a workflow tax of its own.
The third gap is translation. The UI ships in English, Simplified Chinese, and Japanese, but auto-translation of generated guides isn’t there. That’s a missed opportunity for cross-border operations, where the person executing your SOPs often works in a different language than the person writing them. Even a rough machine translation would be better than asking a VA to interpret a screenshot-heavy English guide while trying to click.
Finally, the maintenance problem hasn’t been fully solved. When one commenter asked how a guide holds up after the underlying app UI changes, the maker answered honestly: StepShot does not automatically detect when a third-party app changes its UI. That means your SOPs will still rot. The tool makes the first recording cheap, but it doesn’t watch your marketplace’s interface changes for you.
Where the math breaks
The pricing is deliberately out of step with today’s SaaS norms. Pro at $129 and Team at $599 for five seats, perpetual and all future versions included, looks almost old-fashioned. The free tier is not a trial: capture and editing are unlimited, and exports cover the first 10 steps in Markdown and HTML. For a small task, free is genuinely useful. For a real SOP — say, a 30-step inbound shipment process — you’ll pay.
For a one-person all-Mac operation, $129 is a no-brainer. For a team of five, $599 is competitive with a year of subscription tools. But the real cost isn’t the license; it’s the hardware. Every seat has to be a Mac on macOS 14+. If your VAs use Windows, you have to either buy them Macs or keep the tool away from them. At that point, the math weakens. A documentation tool is only worth what the least privileged person on your team can run.
What I’d watch / test next
Don’t buy five seats on Day 1. Put StepShot on a Mac with no customer data and record one real workflow — say, confirming a Shopify order or creating a shipment in the marketplace admin. Export the Markdown and ask yourself: could a stranger in another country execute it without sending you questions? Then test redaction with fake PII. Put a mock address in a screenshot, run the OCR redaction, zoom into the exported file, and verify the data is gone — destructive should mean destructive. If that passes, take the maker up on the offer in the launch thread for a 30-day Pro key, and record a workflow that touches a native app or a terminal, not just a browser. Then watch for Windows support, team accounts, cloud sync, Notion export, and auto-translation. If Windows ships and sharing improves, this becomes a genuinely serious cross-border onboarding tool. Until then, treat it as a promising local-first experiment with an unusually honest privacy model.
