[bleonard]

There are usual suspects for sources. Databases (MySQL, Postgres) and data warehouse (Snowflake, BigQuery, RedShift).

Though data out is more common, we can also bring in data from any given SaaS tool as well. For example, we have a Mailchimp _source_ that will pull in people as they signup through their form.

There is a plugin model and a few Typescript interfaces to implement to be either a source or a destination.

[psadri]

Is it possible to pipe data through some intermediate processing step? Say to normalize it? Or is it always source directly connected to an output?

[bleonard]

We can use any query to bring in and are considering various ways to process it after that. That's in the near term roadmap.

The main kind of "processing" that's done now is using all these properties to calculate cohort membership (High Value users) so that all these tools can use it: Zendesk to route tickets, Marketo to trigger a campaign, even the product to change their dashboard.