Bright Machines says it now gives customers governed access to factory dashboards without opening inbound connections to its air-gapped plants. Its Databricks-based setup makes production data queryable within one to two minutes, while replacing a fragile nine-service pipeline with a single auditable route out.
Databricks Watch analysis
What happened
Bright Machines captures changes from factory PostgreSQL databases at the edge, writes them as Apache Iceberg data and sends them out through one controlled path. Databricks reads the data in place, with Unity Catalog handling permissions. The company says the first customer-facing dashboard went live about two months after work began.
The Databricks case study describes dashboards that refresh every ten minutes. A heartbeat check flags a broken pipeline in under 30 minutes, rather than leaving failures unnoticed for days. Bright Machines says it replaced nine loosely connected services and kept inbound access to its plants closed.
Why it matters
This is a practical example of the trade-offs behind sharing data from a locked-down industrial site. Rather than letting a cloud service reach into the factory network, Bright Machines pushes data out and retains control over when validated tables become available. Customers can see production information without the company maintaining a separate report-and-email routine for each request.
The architecture also shows where the complexity goes: the company says its design had to account for many small data objects, and that it would build a reconciliation process earlier next time. Safer access is valuable, but keeping the pipeline accurate and observable still takes engineering work.
Our read
The useful lesson is not that every factory needs this exact stack. It is that the security boundary shaped the design from the start: one outbound path, permissions applied to shared data, and a heartbeat to distinguish an idle production line from a failed pipeline. That is a more convincing case study than another dashboard with a shiny coat of paint.
What to watch
- Whether Bright Machines shares more on how the setup performs across additional customer sites.
- How its planned natural-language analytics and AI features use the newly accessible factory data.
- Whether the reconciliation process it says it would build earlier becomes part of the production design.
Discussion spark: For sharing data from a locked-down factory, should the priority be keeping every route auditable, or making the system simple enough that a small engineering team can reliably maintain it?
Sources and evidence
- Air gapped, not data dark: how Bright Machines gives customers live factory dashboards without opening the plant (11 November 2026, 16:10 UTC)
Independent WittyWires tracker for public updates about Databricks. Not affiliated with or endorsed by Databricks; this is not an official account.