Work that runs automatically
Set it up once and it keeps running. Automations respond to data changes, scheduled times, or manual triggers. Code-first but generated via natural language. Powerful without the complexity.
Work that happens on time whether or not anyone remembers
Somebody on your team is doing something every Monday morning. Pulling numbers, sending a digest, reconciling two systems that do not talk. It happens because they remember, and the week they are on holiday it does not happen.
A schedule is that job, running on its own. Any interval you like: every minute, daily at nine, the first of the month. It runs on our machines, so closing a laptop changes nothing.
What makes it worth trusting is what happens when a run goes wrong. A failed run is recorded with its output and its timing, and the schedule keeps its next interval rather than stopping. Missed runs are handled the way you choose: catch up immediately, skip to the next one, or tell somebody. You can pause a job without deleting it, trigger it by hand to test a fix, and read back every run it has ever done.
The thing you are buying is not the cron expression. It is not having to remember.
Work that takes three days and survives all three
Some work does not finish in one go. An order that needs a credit check, then a human to approve the exception, then a shipping label. Onboarding that waits on someone to sign something. The parts are simple and the waiting is what breaks it, because whatever was holding the state has usually gone away by the time the answer comes back.
A workflow holds the state instead. Every completed step is saved, so a restart picks up from the last one rather than starting over. Steps that fail retry on their own, with a backoff you set, which is most of what goes wrong when two systems have to agree.
And it can wait for a person. A workflow pauses on an approval and sits there, for minutes or for days, until somebody approves or rejects. That is the part that lets a real process live here rather than only the parts with no humans in them. You can see every instance at any moment, queued or running or paused or finished, and read back every step and every retry when you need to explain what happened.
Most tools make you choose between a process that is automatic and a process with a person in it. This is the one where waiting for somebody is a step like any other.
What Automations gives you
Event Triggers
React to data changes, app lifecycle events, or custom signals. When X happens, do Y.
Schedule Triggers
Run automations on a cron schedule. Daily cleanups, weekly reports, monthly aggregations.
Manual Triggers
Click a button to run an automation. Perfect for on-demand tasks.
Integration Access
Automations can call any connected integration. Send Slack messages, create HubSpot contacts, charge Stripe customers.
Entity Access
Read and write workspace entities. Update records, create new entries, aggregate data.
Error Handling
Built-in logging, error capture, and retry logic. Know when things fail and why.
Why Automations matters
- Code-first means full flexibility. No visual builder limitations.
- Generated via chat, so non-developers can create them
- Deep integration with workspace entities and connections
- Runs on a global edge network. Fast everywhere.
How Automations works
Automations in Runwork are event-driven functions that execute automatically when specific conditions are met. Unlike visual automation builders that limit you to predefined actions, Runwork automations are code-first, giving you full flexibility while still being easy to create through natural language with your agent.
Every automation starts with a trigger. Event triggers respond to data changes: when a customer is created, when an order status updates, when a specific field changes. Manual triggers let you run automations on-demand with the click of a button.
The automation body is where the logic lives. You have full access to your workspace entities: query customers, update orders, create new records. You can call any connected integration, whether that means sending a Slack message, creating a HubSpot contact, or processing a Stripe charge.
You create automations in your own AI tool. Tell your agent: "When a new customer signs up, send them a welcome email and create a task for the sales team to follow up." Your agent writes the automation, tests it in a sandbox, and deploys it to your team cloud, where it keeps running for everyone. You can review the code, modify it, or have your agent iterate.
Reliability is built in. Automations run on Runwork's global edge network with automatic error handling, retry logic, and comprehensive logging. When something fails, you know exactly what happened and why. Failed automations can be replayed with a single click once the underlying issue is fixed.
Sales, marketing, operations, finance: the work that runs on a schedule and nobody enjoys owning. Describe one and your agent builds it, then it runs in your team cloud where everyone's AI can reach it.
Frequently Asked Questions
What is an automation in Runwork?
How do I create an automation without coding?
What can automations access in my workspace?
How are automations different from workflows?
What happens if an automation fails?
Do we have to build every automation from scratch?
Use Cases
Related Features
See How Teams Use Automations
Ready to try Automations?
One shared cloud under the AI tools your team already uses.