Say what you want automated. Cronable builds it and runs it.

Yes. The Cronable engine installs and runs entirely on your own server or VM, and there is no telemetry of your data. Nothing of yours reaches us unless you switch on one of three optional add-ons: the hosted webhook relay (only if you take inbound webhooks), the hosted AI assistant (only if you pick it over your own AI key or local Claude CLI), and remote control (only if you want the dashboard from outside the machine). Each is a pipe we don’t retain: the assistant sends your messages and what it needs to answer them, and remote control carries your dashboard traffic in transit — including a secret at the moment you type one in remotely. Leave all three off and nothing leaves your infrastructure. Our Privacy Policy sets out each one in full.

For the jobs you build, Cronable runs Claude and other agents on your own machine using your own AI accounts and MCP connections — however you authenticate, with no markup and nothing routed through us. The built-in AI assistant is the one exception you control: point it at your own AI key or your machine’s local Claude CLI (both stay on your box), or use the zero-setup hosted Cronable assistant, which sends your messages and what it needs to answer them — your job configuration and, when you ask about a run, that run’s secret-masked output — through us to a third-party AI provider to generate replies (we meter, but don’t store, those messages). See our Privacy Policy for exactly what is and isn’t sent.

Cronable ships with Claude — both the Claude Code CLI (a coding agent in a repo) and the Anthropic Messages API — and OpenAI, via the Chat Completions API and the Codex coding-agent CLI. Pick the right model and mode for each job, all running under your own AI accounts — however you authenticate — with no vendor lock-in.

Yes — Cronable orchestrates them. Run an OpenClaw or Hermes agent as a step in a pipeline, reuse the skills they already have, or drive Cronable from their own chat over its MCP server. They’re great for authoring a job by chatting; Cronable is where you operate it — on a visual canvas, with schedules, retries and the full history of every run. Chat to author, canvas to operate.

Cron and launchd fire each job on a timer, blind to everything else — if the 3am export fails, the 3:05 job still runs, on whatever broken state it left behind. Cronable turns those isolated timers into a dependency graph: a step runs only when the one it depends on actually succeeded, failures branch to a handler instead of cascading, results flow from one step to the next, and you watch the whole thing on a live colour-coded canvas with per-run logs. Same one-file-per-job simplicity — none of the silent breakage.

A chat agent does the work but hides the state: to find out what’s scheduled or whether last night’s run failed, you have to ask it and trust the answer. Cronable makes the automation the thing you look at — a live, colour-coded canvas of every job, with the full secret-masked log of each run a click away. It also does what a single scheduled prompt can’t: run long chains and branching trees where each step fires only if its parent succeeded, failures route to a handler, and data flows from one step to the next. And plenty of jobs have no AI in them at all — a reminder, an email, an HTTP call, a database query. AI is one option, not the point.

Three ways. Drag and drop on the visual DAG editor in the browser; describe what you want to the built-in AI assistant and it creates the job right in the console; or connect Claude or ChatGPT to Cronable’s MCP server and ask from there. The built-in AI assistant is on every plan — including Free — with a daily message limit that scales as you move up; the MCP server is on Business. However it’s made, the job lands in your own git repository, and every AI change runs under the same roles, approvals and audit as a hand edit.

Yes. Triggers start pipelines from their own source: an inbound webhook, a file landing in a watched folder, new mail in an IMAP inbox, an upcoming calendar event, a new message to your Telegram bot, a new row in a Google Sheet — and more. Webhooks arrive through Cronable’s hosted relay over a connection your server opens outbound, and every other trigger polls outward — so no inbound ports or public URLs are ever required, even behind NAT or a firewall.

Only the inbound event itself, and only if you use webhooks. Cronable’s daemon never opens a port; it holds an outbound connection to our relay, and third parties post to a public relay URL that forwards the event down to your server — we don’t store it. Nothing else leaves on account of webhooks: your jobs, secrets, credentials, AI accounts and every command Cronable runs stay put. (Two other features can send content through us if you turn them on — the hosted AI assistant and remote control, which carries your dashboard traffic in transit. Both are opt-in and covered in our Privacy Policy.) Want zero cloud touchpoints? The polling triggers (email, calendar, folders, Telegram, Sheets) reach out from your server and never touch the relay.

Secrets and typed credentials (API keys, bearer/basic/header auth, database connection strings) are AES-256-GCM encrypted at rest, never written to git and masked from logs. The web UI is authenticated with per-user logins — or your identity provider via OIDC single sign-on — with brute-force lockout, and every login, secret change and settings edit lands in an append-only audit trail. The engine binds to localhost and runs as a dedicated low-privilege user.

Yes. Extensions are self-contained modules — one folder each — so your team, or ours, can add custom job types that behave exactly like the built-ins.

Yes. The database job type runs SQL against Postgres, MySQL or SQL Server using a stored connection-string credential, with proper bind parameters and row limits. Results flow downstream as JSON, so a query can feed an AI summary, a report email or any other job in the chain.

Simple self-serve tiers billed monthly via Stripe: Basic, Plus and Business, metered on jobs, inbound webhooks and users — git sync and encrypted secrets are on every plan. Enterprise is custom — contact us. Cancel anytime. See the pricing page for current rates.

Three ways. Install the Mac or Windows desktop app and double-click — it runs in your menu bar or system tray and manages the engine for you, no terminal needed. Or, for an always-on box, install a headless server on macOS or Linux from the command line and operate it from anywhere through the same dashboard via remote control (your server dials out, so there are no ports to open). Either way there is no manual configuration.

If you want. Self-install if you prefer to run it yourself — or we come to you: we install everything, run a workshop to map what to automate, build the workflows with you, and train your team to run and extend them on their own. Available as a setup add-on on Plus and Business, and included with Enterprise.

Yes. Signed updates roll out automatically in the background with rollback, on every plan. Community support + docs come with all plans; Plus and Business add email support; Enterprise includes a dedicated SLA.