Scheduled AI workflows · Alpha
AI that works on your clock.
Design a workflow once. cronai.app runs it on schedule, on the models you choose, and keeps a record of every run — what ran, when, and how it ended.
| Started | Step | Detail | Took |
|---|---|---|---|
| 07:00:00 | Trigger | 0 7 * * 1-5 | – |
| 07:00:01 | AI step | gpt-5 · web search | 29 s |
| 07:00:30 | Deliver | 1 s |
Completedin 30 s, delivered to 1 recipient
How it works
From prompt to record in four steps.
01
Design the flow
Write the prompt, pick the model, choose where the result goes.
02
Set the schedule
A cron expression decides when. The playground below shows exactly when that is.
03
It runs
When the schedule fires, cronai.app runs the flow exactly as you designed it. The record shows when the run actually started.
04
See the outcome
The result is delivered, and the run is in the history with its status, timing and output.
Principles
Not autonomous. Accountable.
Plenty of tools promise agents that decide for you. cronai.app does the opposite: a person designs the flow, and the machine follows it — on time, and on the record.
- You design the flow
- cronai.app runs the flow you defined, as you defined it. It does not invent a plan of its own.
- Every run leaves a record
- When it started, how long it took, what came back and how it ended — kept for every run, not only the failures.
- Failure is recorded, not hidden
- A failed run shows up as failed, with the error that stopped it.
- You choose the model
- Every AI step names the model it runs on. Nothing is swapped in behind your back.
Capabilities
What it does today, and what comes next.
Every capability carries its state. A roadmap item is not a shipped feature, and this page says which is which.
In the alpha
Cron schedules
Standard five-field cron expressions, from every minute to once a year.
In the alphaAI steps
Each run sends your prompt to the model you picked. A step can search the web for current information if you choose a search provider for it.
In the alphaDelivery
Results arrive by email, or as a webhook to the tools you already use.
In the alphaRun history
Every run is recorded with its status, timing and output.
In the alpha
Coming next
Multi-step workflows
In developmentWorkflows will chain steps, branch on a result, run steps in parallel and pass typed outputs to the next step, and every change will be kept as a new version.
In developmentMore model providers
In developmentA step will be able to reach models from more providers through a gateway.
- OpenRouterIn development
- Fallback across providersPlanned
- OrcaRouterPlanned
Guardrails
In developmentUsage quotas per organisation will keep a schedule from running away with your budget, and checks on the output will follow.
- Usage quotasIn development
- Output checksPlanned
Human handoff
PlannedAn approval step will hold a result until a person has looked at it.
Planned
Where it's heading
A person sets the bounds: the time window, the budget, the models a job may use. Inside them, the scheduler will decide the details, and every run will still end in a record.
Adaptive timing
A schedule will be able to run inside a window you set, starting when its inputs exist and the provider is answering.
PlannedAdaptive retry
Retries will respect the provider's Retry-After and the job's own history, and stop with a reason when a retry cannot help.
PlannedContext between runs
A workflow will be able to keep a small memory between runs that you can read and clear.
PlannedYour own models and infrastructure
Steps will be able to use your own model keys and connect to your own infrastructure.
Planned
Beyond cron: what a scheduler for AI workflows has to do (opens in a new tab)
Playground
Read a schedule on your clock.
Pick a use case or type any cron expression. The next runs appear in your own time zone.
Five fields: minute, hour, day of month, month, day of week. Shortcuts such as @daily work too.
Morning briefing
In the alphaA summary of overnight news on your topics, in your inbox before the first meeting.
Delivered by email
Working out the next runs on your clock.
FAQ
Questions, answered plainly.
What is cronai.app?
A service for scheduled AI workflows. You describe a workflow once — the prompt, the model, where the result goes — and a cron schedule decides when it runs. Every run is recorded, so you can see what ran, when, and how it ended.
What does alpha mean here?
It works, and it is unfinished. Access is by invitation from the waitlist and for business use only, things change between versions, and anything on this page marked in development or planned is not available yet.
Which models can I use?
OpenAI models in the alpha; a step can also search the web. More providers through model gateways are in development, starting with OpenRouter; OrcaRouter is planned.
Does it act on its own?
Only in the sense that it runs on schedule without you pressing a button. It follows the flow you designed and does not decide what to do next. That is the point.
What does it cost?
The alpha is free of charge. There's no pricing yet. Alpha terms
Where is my data processed?
By default in the European Union: the app and its database run in Frankfurt, and AI steps use inference in the European Union. E-mails the app sends are processed in the United States, and options that are processed there say so where you choose them. Privacy
Who builds it?
CronAI UG (haftungsbeschränkt), a founder-led software company in Munich. cronai.app is one of the products CronAI builds and runs itself. More about CronAI
Where are the docs?
On three pages. Start with cronai.app 101, which takes you from your invitation to your first workflow and its runs. The API reference is for calling the API: its operations by resource and how to authenticate. How schedules run is for knowing when a workflow runs: cron expressions, time zones, clock changes, and what happens when a run fails or a time is missed.
Put your first workflow on the clock.
Join the waitlist to get an invitation to the alpha.