AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Read, don't copy: replacing file and API feeds with OpenSharing - Part 1 (The Why)

Collected Nov 6, 2026

Databricks has published the first part of a series making the case for OpenSharing as an alternative to the file- and API-based feeds many teams currently use to ingest customer data. The post is framed around organizations that pull data from customers over SFTP or APIs, and it sets out the argument for reading shared data rather than copying it. The second part of the series, covering the how, is not described in the excerpt.

OpenSharing is Databricks' open protocol for exchanging data and AI assets across platforms and organizations, so consumers access a shared dataset where it lives instead of receiving a duplicated copy. In practice, that distinction is the point of the piece: an SFTP drop or a custom API feed is a coupling between producer and consumer, with each side owning a piece of pipeline that has to be scheduled, monitored and renegotiated when schemas shift. A sharing relationship moves that contract to a declared share.

For developers and product teams building ingestion, this suggests effort shifts from file and endpoint plumbing toward defining what is shared, with whom, and under what access controls — an inference, since the details of the approach are presumably the subject of the upcoming part.

Why it matters: teams that maintain bespoke SFTP and API ingestion pipelines for customer data are the audience most directly affected, and the argument implies replacing copy-based feeds with read-based shares reduces the surface area they have to operate. Whether that trade-off holds for latency, governance and cross-organization access is likely to depend on implementation details the series has yet to set out.

Read at Databricks

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

If you ingest data from your customers over SFTP or APIs, you have probably asked...