Events, alerts and monitoring
The world changes.Your workflow starts.
Almost every automation waits for something in your own database. This one can wait for something outside it: a festival announced in a city where you have three venues, a severe weather warning over a region where your crews are working, an outage at a provider you depend on.
What is in the feed.
A live worldwide events feed built from more than 175 provider adapters and refreshed every ten to twenty minutes. It holds two kinds of thing that are usually sold separately.
Things people go to: concerts, festivals, expos, trade fairs, sports fixtures, conferences.
Things people react to: earthquakes, flood and weather warnings, air-quality bands, aviation delays, cloud and service outages, product recalls, public tenders.
Every record carries a venue, a city, a country and coordinates; a time in UTC and in the venue's own timezone; translations where the source provided them; a source URL; and its licence and attribution. Price is free, paid, or honestly unknown — the third of those is a deliberate value rather than a gap, because a feed that guesses at prices is a feed you cannot act on.
The roofing firm and the weather.
A company with fourteen crews working outdoors across southern Poland. When a severe warning goes up, the crew leads need to know before they set off — which means before six in the morning, which means not by somebody watching a website.
The workflow is short. The events feed fires on a severe alert in Poland. A condition asks whether any crews are inside the affected area, using coordinates from reference data. If so, quiet hours are overridden — deliberately, because this is one of the few alerts that is supposed to wake people — and each crew lead gets an SMS.
Everything is recorded: which alert, from which source, with the source URL, so the conversation afterwards about whether it was the right call is a conversation about facts.
Why the timezone matters.
Because an event's own timezone and the recipient's timezone are different questions, and a system that conflates them sends the reminder at the wrong hour to somebody in another country. Both are present on every record, so a workflow can be right about both.
- ⚡TriggerCategory
alert, severity Severe, country PL. - 2Area matchedVoivodeship boundary against crew locations.
- ?ConditionCrews inside? Yes — 14 of them.
- !Quiet hoursOverridden for severity Severe. This one wakes people on purpose.
- ✉SMS14 messages. Delivery states tracked back onto the run.
- 6FiledSource URL and attribution recorded with the run.
175+ sourcesRefreshed every 10–20 min
Something happens somewhere.
Your workflow already knows.
The other shapes
Six more ways this gets used.
Demand you can see coming
A hotel, a restaurant or a car park in a city where a three-day expo has just been scheduled. Staffing and pricing decided six weeks out rather than on the Friday.
Timely content
A post or an offer that refers to something actually happening locally. Marketing that is timely beats marketing that is clever, and timeliness is a scheduling problem.
Supplier outages
A provider you depend on reports an incident, and the workflow that would have called it pauses itself instead of generating four hundred failures.
Air quality
A band crossed in a city where you run outdoor work or have vulnerable customers, and a message goes out with the actual number in it.
Recalls
A recall published for a product category you stock, surfaced against your own catalogue, with the source so somebody can check it properly.
Tenders
Public procurement notices matching what you actually do, filtered by an AI step reading the description rather than by keywords that miss the half worth bidding on.
Honest limits.
In alerting, dishonesty hurts.
The feed refreshes every ten to twenty minutes. That is fast enough for weather, outages, recalls and announcements. It is not a life-safety system, and nothing here should be the only channel by which somebody learns that a river is rising. Use it alongside official warnings, not instead of them.
Coverage is broad but not uniform: some countries and categories have several sources and some have one. Where a source is thin, that is visible on the record rather than smoothed over. Price is unknown far more often than people expect, and we would rather show you that than invent a number.
Attribution is not optional. Every event carries the licence it was published under and the attribution that licence requires. If a workflow puts an event in front of the public — on a page, in a newsletter, on a screen in a foyer — the credit travels with it. That is a term of using the data, not a nicety, and it is handled for you precisely so it does not get forgotten.