Skip to main content

​
Overview

Data sources are the raw inputs to ProDex — the files and systems that carry the operational data your work builds on. There are two ingestion paths: file uploads, which you drive yourself, and system integrations, which your ProDex team provisions against your operational systems. Sources are the raw material; ProDex does not modify them. Once a source is in ProDex, Dexter reads and profiles it, resolves its codes and quirks with you through Exploratory Analysis, and turns it into durable artifacts: Insights dashboards you revisit, pipelines that document how numbers were derived, and inputs to simulation and planning models. You manage sources from Data in the left sidebar, under the My Data tab. For how sources fit into the wider data lifecycle, see Data Overview.

​
Get Started

Upload a file in any conversation — drag it into the chat, or use the upload control on the Data page. Once it’s in, ask Dexter to explore it: the Exploratory Analysis workflow profiles the file, resolves ambiguous codes with you, and produces a first dashboard plus the pipeline behind it. From there the same source can feed simulation models, planning models, BOMs, and reports. For integrations, contact your ProDex representative — connectivity to your operational systems is scoped and provisioned with your team during onboarding.

​
File Uploads

The self-serve path. Drop a file into any conversation with Dexter (an MES export, an ERP report, a spreadsheet, a work-instruction PDF, a CAD file), or use the upload control on the Data page. Uploaded files are the fastest way to get real data in front of Dexter without waiting on IT. Supported formats: Individual files can be up to 2 GB.
A structured export always beats a screenshot of one. Dexter can read charts and tables in images, but exact figures — especially from dense dashboards — are far more reliable from a CSV or Excel export when one is available.
Where uploads live. In your factory workspace. Files persist across conversations and are available to anyone with access to the Factory. You can review and remove uploaded files from the Data page.
Uploaded files are factory-scoped, not user-scoped. Anyone with access to the factory can read any file uploaded to it — there is no per-user privacy boundary on uploads. Don’t upload material to a factory whose access list isn’t appropriate for it.

​
System Integrations

The managed path. An integration links Dexter to one of your operational systems (ERP, MES, WMS, data warehouse, IoT historian, and so on), set up with your ProDex team during onboarding. Part of that setup is a per-system documentation record — administered from the Admin panel’s Integration Documentation editor — that tells Dexter the shape and semantics of that system’s data, so it can interpret exports and answer questions about the source with the same context every time.
Your data stays in your systems. Access is governed by your team, and ProDex does not replicate the underlying data unless a pipeline materializes a specific extract.
ProDex works with data from systems across the manufacturing stack. The set below is representative — reach out about your specific systems, and your ProDex team scopes the right setup: ERP
MES
WMS
BI
Cloud Platforms
Data Warehouses and Lakehouses
Databases
Industrial and IoT
CRM
Salesforce (for demand signals)
Messaging
Don’t see your system? Contact us — the integration surface is extensible. How it works with Dexter. With an integration set up, Dexter draws on its documentation record to understand how that system is structured, so exports from it profile faster and derivations built on a pipeline start from settled context rather than raw guesswork. Connection setup. Scoped and provisioned by your ProDex team during onboarding — including the documentation record Dexter reads. Request an integration through your ProDex representative. Security. Whatever access an integration involves is governed by your team, scoped to your factory, and not shared with other factories in your organization.

​
What Dexter Does with a Source

Whether the source is an uploaded file or an integration, the flow is the same:
  1. Profile. Dexter reads the full source when practical rather than sampling, documents its schema and value distributions, and flags anomalies.
  2. Interpret with you. Dexter surfaces codes, tags, and terms it isn’t certain about and confirms interpretation with you before extracting from them.
  3. Land in durable artifacts. The analysis produces an Insights dashboard and a pipeline that documents the derivation — every number on the dashboard traces back through the pipeline to the source.
  4. Feed downstream work. The same source, through the same or a new pipeline, can supply parameters to a simulation model, quantities on a BOM, inputs to a planning model, or figures in a report.
The confirmed interpretation of a source — its schema, codes, units, and quality caveats — is captured in the source node’s description on any pipeline that reads it, so future work on the same source starts from settled ground.
Every source surfaces the same way in a pipeline — as a source node the derivation reads from — so the work you build on top is expressed against that node, not against the raw file’s storage details.

​
After a Source Is In

With a source connected, you can:
  • Run Exploratory Analysis to produce a dashboard and a pipeline
  • Feed simulation models with fitted parameters (arrival rates, cycle times, yields) traceable back to the source
  • Populate BOMs and planning models with real quantities and lead times
  • Generate reports whose figures link back to the source through their pipeline
Every one of these is authored by Dexter with the source as an input node in the pipeline that produced it, so downstream users can click from a number back to the file or integration it came from.