Pipeline integrations
The Integrations page of a pipeline, with the input integrations that start or feed it, the output integrations that receive its rows, and the alert rules that watch its runs and data.
Integrations is the entry between Inputs and Changelog in the pipeline’s How it works rail. It gathers every connection that touches this one pipeline: what starts it or brings data in, where its rows go, and who hears about its runs. The same integrations and rules appear on the workspace’s Integrations page; this view filters them to the pipeline you are looking at.
- Input integrationsstart or feed it
- Pipelineruns and publishes
- Output integrationsreceive its rows
- Alert rulestell people what happened
Open a pipeline, choose How it works, then Integrations in the rail. The link is #integrations.
Input integrations
What starts this pipeline or pulls data in for it.
An integration is listed here when it is a trigger for this pipeline (a webhook that starts a run, with or without inputs), or a source that fills a dataset this pipeline reads or publishes (a Google Sheet, a warehouse query, an S3 bucket, a JSON API). Each card shows the service, the integration’s name, its role and its status: Live, Needs setup, Requested or Revoked. Click a card to open the integration’s page. The card menu has Delete, which asks Delete "<name>"? This can't be undone.
The last tile is Add input. It opens Connect Input, a list of the services that can start or feed a pipeline:
| Service | What it does |
|---|---|
| Google Sheets | Send results to a sheet, or bring rows in from one. |
| Snowflake | Keep a Snowflake table in sync, or bring a query result in. |
| S3 | Write rows to a bucket, or read objects in. Any S3-compatible store. |
| Postgres | Keep a Postgres table in sync, or bring a query result in. |
| MySQL | Keep a MySQL table in sync, or bring a query result in. |
| BigQuery | Keep a BigQuery table in sync, or bring a query result in. |
| Webhooks | Send rows or alerts to your URL, or receive JSON from your system. |
| Read from an API | Pull rows from a JSON URL into a dataset. |
Filter services… narrows the list. When nothing matches, Ask Jason about “…” offers to set up something that is not listed.
Picking a service opens Connect <service>, with the note Jason sets this up. You add any keys yourself. Describe what the integration should do in What should it do? (suggestion pills help) and press Send to Jason. Jason opens with a new chat that holds your request and names this pipeline, sets the integration up, and links you to the integration page for anything only you should enter, such as a key or a token.
Start this pipeline whenever our catalogue system posts a new product list, and use the posted list as the product input.
Triggers are covered in Triggers; warehouse and storage connections in Warehouses and storage.
Output integrations
Where this pipeline’s rows are sent or published.
An integration is listed here when it receives the rows of a dataset this pipeline publishes: a destination such as a sheet, a warehouse table, a bucket or a webhook, or a channel that receives new rows through one of this pipeline’s alert rules. The cards work as above.
The last tile is Add output, which opens Connect Output:
| Service | What it does |
|---|---|
| Google Sheets, Snowflake, S3, Postgres, MySQL, BigQuery | As above, in the sending direction. |
| Email alerts when runs fail or data changes. | |
| Slack alerts | Post alerts and new rows to a Slack channel. |
| Browser notifications | Browser alerts, even when Jsonify is closed. |
| Webhooks | Send rows or alerts to your URL, or receive JSON from your system. |
| Publish an API | Serve a dataset’s latest rows at a GET endpoint. |
The rest of the flow is the same: pick a service, say what it should do, Send to Jason. See Destinations and Warehouses and storage.
Keep a Snowflake table in sync with Live product offers, upserting on retailer and sku after every run.
Alert rules
Who hears about this pipeline’s runs and data, and when.
Every alert rule that watches this pipeline’s runs, or a dataset it reads or publishes, is a card here. Each card has an on/off switch that saves at once, and Delete in its menu (Delete "<name>"? This can't be undone.). Click a card to open the rule.
The last tile is Add alert rule. It opens the new-rule dialog with this pipeline already chosen. If the channel you want does not exist yet, the dialog offers to add an integration, which takes you to the workspace’s Integrations page. Every rule option is described in Alert rules.
Where else these appear
An integration is not part of a pipeline version. Adding or removing one never creates a new version, and a restore of an older version keeps the integrations you have today.