Data Sources

​

Overview

Wherever your data lives, ProDex can ingest it. The platform supports two primary ingestion paths: file uploads (the self-serve path — drop an MES export, ERP report, or spreadsheet into a conversation) and system integrations (the managed path — direct connections to your databases, warehouses, ERPs, MES, and more, provisioned by your ProDex team). Dexter — ProDex’s AI assistant — is what reads, profiles, and works with your data once it’s in the platform. Whether it’s a column you uploaded or a table from your warehouse, Dexter inspects it, asks clarifying questions, and helps you decide what to derive from it. For system integrations, your data stays in the systems your organization already owns and governs — ProDex reads from those systems under credentials your team controls, and source data isn’t replicated unless a pipeline materializes it. Files you upload to a conversation are stored in your factory’s workspace and persist across conversations and sessions.

​

Get Started

The fastest way to begin is to upload a file in any conversation: drag a CSV, Excel, or PDF into Dexter’s chat panel, and Dexter will profile it, ask clarifying questions, and help you decide what to build from it — no integration setup required. From there you can build a Pipeline that turns the upload into reusable simulation, planning, or report inputs.

​

File Uploads

The self-serve ingestion path. Upload files directly to a conversation or via the Upload button on the Insights page; uploaded files persist at the factory level and are available across conversations. Supported file formats:

• Spreadsheets — CSV, XLSX, XLS, XLSM, XLSB
• Documents — PDF, DOCX, PPTX
• Plain text and structured text — TXT, Markdown, JSON
• Images — PNG, JPEG (Dexter reads charts, dashboards, and tables in screenshots, though structured exports are more accurate when available)

For the full list of attachment behaviors and how Dexter handles each format, see File Attachments. Common uses for file uploads:

• Spreadsheet or CSV export — download from a source system and attach. Dexter profiles columns and asks clarifying questions before using the data.
• PDF reports and documents — Dexter extracts values, tables, and context.
• Screenshots of dashboards — useful when the data lives in a BI tool view that’s cumbersome to export. Screenshots work for extracting approximate values, but exact figures (especially from dense charts) are better served by a structured export when one is available.

​

System Integrations

The managed ingestion path. Direct connections to systems your organization owns — provisioned by your ProDex team, then queryable through Dexter and pipelines. The systems below have direct integrations. Setup is handled by your ProDex team — see How Connections Are Set Up below. If a system you need isn’t on this list, the File Uploads path covers most cases as a starting point. ERP MES WMS BI Cloud Platforms Data Warehouses and Lakehouses Databases IoT and Shop Floor CRM Messaging Don’t see your system? Contact us — we can add it.

​

How Sources Show Up in Pipelines

Both file uploads and system integrations surface as source nodes in a pipeline template. The same pipeline structure works for both — only the source’s binding differs (an uploaded file vs. an integration extract). This means a pipeline you build today against an uploaded file can later be re-pointed at an integration source when the connection is provisioned, without rebuilding the rest of the pipeline.

​

What Happens After Data Enters

Once data is in the platform, you can:

• Build a Pipeline — transform the data into reusable, auditable simulation parameters, planning inputs, or report outputs
• Run Experiments — compare what-if scenarios across multiple Snapshots that draw on the data
• Feed it into Production Planning, BOMs, or the Modeler — pipeline outputs land directly on those pages

​

Data Flow

​

How Dexter Uses Connected Data

Once a connection exists, Dexter can:

• Query live — ask questions against source data directly (“What’s our average cycle time for Product X across the last 30 days?”) without waiting for an export
• Profile and document — inspect schemas, value distributions, and data quality, then document what it finds before you build on top
• Build pipelines on top — use connected sources as nodes in a pipeline, deriving simulation parameters from live data with full traceability
• Schedule recurring jobs — set up a scheduled job that kicks off a pipeline run on a cadence you define. Each scheduled run creates a conversation you can inspect; the run uses whatever inputs are bound to the pipeline at trigger time.

Every query Dexter runs is visible in the conversation, so you can audit exactly what data was read and what was done with it.

​

How Connections Are Set Up

ProDex operates a managed-integration model today: your ProDex team provisions connections during onboarding and adds new ones on request. Most integrations are live within a few days of onboarding; complex setups — VPN tunnels, custom schema mappings, write-back paths — may take longer. Provisioning covers credential setup, network configuration, schema mapping for your key tables, and any transformation logic required to normalize source data into the shapes Dexter expects. To request a new connection, contact your ProDex representative with:

• The system you want to connect (name and version if relevant)
• The kind of data you want ProDex to access (tables, schemas, or specific data domains)
• Any security or compliance constraints your IT team requires (read-only, VPN, IP allowlist, etc.)

​

Security and Access

Integration connections. Connections use credentials provided by your team. ProDex accesses data at the permission level you grant — typically read-only on specific schemas or tables. All data transfer is encrypted in transit. Connection credentials are stored in ProDex’s managed secrets store and are scoped to your factory. If your organization requires specific network controls — IP allowlisting, private links, VPN, or proxy routing — your ProDex team configures those during setup. Uploaded files. Files you upload to a conversation are stored at the factory level. They persist across conversations and sessions until deleted; you can manage uploaded files through Settings → AI Assistant settings in the chat panel.