Business process automation tools, judged by a question none of them want you to ask.

Because every other axis trends in the same direction: more integrations, more AI, more dashboards, more features. That is the marketing axis. The buyer axis is different. Six weeks in, you have 8 to 20 processes encoded inside the vendor you chose. The switching cost compounds weekly. Nobody writes the 'great tools I migrated off' blog post because migrating off most BPA tools is a rebuild, not an export. Removability is the axis the marketing pages never raise because the answer is never in the vendor's favor. Clone is the only tool in the category whose own architecture diagram concedes the point up front.

Open src/components/architecture.tsx. Line 84 reads: 'Remove Clone and your business still runs, your data is still where it was.' Lines 123-130 render a legend with two color keys: the first is a small indigo square labeled 'Clone layer (removable)' at line 127; the second is a stone square labeled 'Your existing stack' at line 129. The legend sits directly underneath the architecture stack diagram, so every visitor to the homepage's architecture section sees the removable label next to the product they are considering buying. No other BPA tool's marketing site does this.

Correct, and that is the point. Open src/components/features.tsx. Across six feature cards, the named apps are QuickBooks, FreshBooks, tl;dv, Fireflies, Otter, Zoom, Google Sheets, Notion, Gmail, HubSpot, Pipedrive, Timely, Calendly. None of these are Clone's products. Clone is the agent that drives them. The category name 'business process automation tool' is usually a euphemism for 'vendor layer in the middle of your stack'. Clone is a BPA tool that refuses that shape.

Each recurring automation is a markdown file at ~/.clone/memory/rituals/<name>.md. The file has a schedule line (e.g. schedule: mon 08:00 local), a plain English description, optional voice notes, and optional stop conditions. The file lives on your Mac. You can open it in any text editor, grep it, diff it in git, copy it to another Mac, email it to a colleague. Uninstall Clone tomorrow and the file still exists and still describes what you were trying to do. That is a different shape from a vendor database row.

Cancel Zapier: every Zap stops firing; invoicing, CRM writes, Slack alerts, lead routing all freeze. Cancel Make: same, for every scenario. Cancel Power Automate: every flow stranded; if you leave Microsoft 365, the flows do not come with you. Cancel UiPath: robots freeze mid-process, selectors waste. Cancel Appian or Pega: processes in vendor XML, migration is a rewrite. Cancel Clone: your markdown ritual files stop being automated, but they still describe, in plain English, what you wanted done. Your QuickBooks, Gmail, HubSpot, and Zoom still open normally the next morning. Nothing is stranded, because Clone was never in the middle of the stack to begin with.

One, loosely. If you build up a long library of rituals, the markdown format, the schedule syntax, and the Planner's conventions are Clone-specific. But the files are plaintext, and every instruction inside them is also plain English. A human rereading your ritual file would understand what the ritual was supposed to do, even without Clone running. Compared to a Zap JSON blob that only Zapier can parse, the asymmetry is stark. You could read your ritual files aloud. They are written for you, not for a vendor parser.

Copilots that draft workflows into an existing vendor canvas are still outputting artifacts locked inside that vendor. The graph is still inside Zapier, the flow is still inside M365, the scenario is still inside Make. The copilot makes authoring faster; it does not change who owns the output. Clone's approach is different: the artifact is a markdown file on your Mac, and the execution layer is an agent that drives your actual apps. There is nothing inside a vendor to author into.

It is a BPA tool in the honest sense: a single product that automates business processes across your stack. What it is not is a graph editor with an execution runtime attached, which is what the category has historically meant. For comparison: a $3,000 to $6,000 per month virtual assistant runs during business hours, can refuse an instruction, gets sick, takes vacation. Clone is $49 per month, runs 24/7, reads the screen, and drives the apps directly. The shape is closer to a programmable assistant than to Zapier, which is why the lock-in question has a different answer.

Three questions. One: if I cancel this tomorrow, what stops running? If the answer is 'lots of things', the tool owns your processes. Two: where are my processes encoded? If inside a vendor database, they are locked. If on my own disk as plaintext, they are not. Three: does the vendor publish a 'migrate off us' docs page that takes less than five minutes to read? If not, assume a lock. Run each tool on your shortlist through these. The list gets short quickly.

Count wise, yes. Meaning wise, the count is not the metric. Clone does not ship connectors because it uses the Computer Agent layer (see architecture.tsx) to read the screen and drive apps directly. That includes apps without any public API, like QuickBooks Desktop, small-firm legal billing tools, your bespoke client portal, legacy ERP. Zapier's 8,000 is the ceiling of what it can ever automate. Clone's zero is shorthand for 'whatever runs on your Mac counts'. Different axis, different answer.

That is the whole appeal of a tool designed to be removable. Install Clone, pick one recurring process (Monday invoicing, Friday admin, Thursday pipeline update), talk to Clone in plain English, watch it drive your actual apps, save the instruction as a ritual. If after 21 days it has not earned its $49, uninstall it. Your apps open the next morning exactly as they did before. Nothing to migrate, nothing to rebuild, no sunset plan. The product is designed so the exit cost is the same as the entry cost: a Command-Q away.

Solo consultants, small advisory firms, boutique accounting, law, and design practices of 1 to 10 people. Teams where there is no dedicated automation engineer, no BPM admin, no RevOps team. Teams whose operators already own QuickBooks, Gmail, a CRM, a transcript tool, and Sheets, and whose problem is not 'we need more tools', it is 'the tools we already bought do not talk to each other and I do it all by hand'. Clone is the BPA tool that takes the stack you have and makes it run itself, without asking you to consolidate, migrate, or rebuild.