Skip to main content

Overview

This is a walkthrough for your first real session in ProDex — the kind of work that fits in one focused sitting. The spine is simple: upload a file, talk to Dexter, get an artifact you can use. Most users’ first productive action isn’t building a simulation — it’s handing Dexter a spreadsheet and asking a real question about it. That’s what this page walks through. Building a simulation model is covered separately in Your First Simulation — come back here first.

Before You Start

You need two things:
  • You’re signed in to ProDex (First Login)
  • A Factory is selected in the factory selector in the left sidebar. If you don’t have one yet, create one from the factory selector — a name reflecting the operation you’re modeling is enough.
You land on Chat by default. That’s where the rest of this walkthrough happens.

Step 1: Attach a File in Chat

Take a real file from your operation — a production log, a cycle-time spreadsheet, an MES export, an ERP demand report, a PDF work instruction, even a screenshot of a report.
Anything close to your actual work is better than a toy example. Real column names, real status codes, and real quirks are exactly what Dexter is built to handle — and what makes the output immediately useful.
Two ways to attach it to a message:
  • Drag the file onto the Chat window — the composer accepts the drop
  • Use the attach affordance in the message composer (next to the text input) to browse and select a file
Supported inputs cover the formats manufacturing teams typically have on hand: CSV, Excel (.xlsx / .xls), PDF, Word, PowerPoint, images, JSON, and CAD files (DWG, DXF, STEP) — see the CAD overview for what happens with drawings. If a format isn’t accepted, Dexter will tell you in the reply. Attach the file, add a short message describing what it is and what you want, and send:
“This is our last 90 days of cycle times from Line 3. Give me the distribution per station and flag anything that looks wrong.”

Step 2: Let Dexter Profile It

On upload, Dexter will:
  1. Profile the file: open it, understand its structure, identify columns, sample values, and get a sense of what each field means.
  2. Ask clarifying questions: usually about column meanings, units, timezones, or the status codes and abbreviations only your team knows. Answer plainly and specifically — “STN_04 is the deburr station; times are in seconds; SCRAP means the part was scrapped downstream, not at that station” is exactly the kind of answer that unlocks good work. Vague answers here produce vague analyses later.
  3. Show its work in the task list: you’ll see the steps Dexter is executing while it works. Profiling a typical spreadsheet takes seconds; larger files or PDFs may take a minute or two.
You don’t need to wait passively — Dexter surfaces anything it needs from you as it goes.
Your answers aren’t throwaway. What you teach Dexter about your data and your operation can be saved to the Knowledge Base — ask Dexter to record it, and the same questions won’t come up again on the next upload.

Step 3: Ask for Something Valuable

Once the file is profiled, ask for something useful. Good first asks:
  • “Give me the cycle time distribution per station and flag any outliers.”
  • “Turn this into a reusable set of processing times I can use in a simulation later.”
  • “Show me demand by SKU by week and highlight the top 10.”
  • “Compare this month’s throughput against the previous three.”
For work with real consequences, Dexter pauses at review briefs — structured summaries with an Approve / Request Changes decision. A brief typically shows the assumptions Dexter made, the numbers it derived, and what happens if you approve. If something is wrong, choose Request Changes and say what needs to change; Dexter reworks and re-presents.
Read the review briefs. They are your leverage: catching a wrong assumption in a brief is far cheaper than catching it later in a downstream plan or model.

Step 4: Find Your Artifact

Depending on what you asked for, the result lands in one of a few places — all inside your active Factory:
  • A pipeline under the sidebar’s Data surface: a reusable, auditable transformation from your uploaded file into derived inputs (cycle time distributions, processing times, demand profiles). Pipelines can be re-run when new data arrives.
  • An insight: a dashboard, chart, or KPI view for exploratory analysis. Insights are factory-scoped and stay available for your whole team.
  • A downloadable report: PDF, PowerPoint, Word, or Excel, when you asked for something shareable.
Dexter tells you where the artifact lives and usually surfaces an action card in Chat with a button that opens it directly — the most reliable way to jump straight to it. Every artifact is yours to reuse — a pipeline feeds a simulation model, an insight anchors a team discussion, a report goes to your executives.

Step 5: When to Graduate to a Project

A plain Chat conversation is the right container when the work fits in one sitting: profile a file, get a chart, draft a report, done. Graduate to a Project when the work will span multiple conversations or needs staged sign-offs — a full simulation study, a multi-week model build, an ongoing planning effort. Projects give the work a durable home with phases, gated sign-offs at each phase boundary, and the ability for Dexter to resume across sessions without you restating everything. A new conversation opens with quick-action cards grouped under category chips — Projects, Quick tasks, and Chat. The New simulation project card under Projects (“Create a project and build a sim from your data”) is the fast path to a scaffolded simulation Project from your uploaded data. Other cards under Quick tasks offer common starting shapes; browse them, or just tell Dexter what you’re trying to do. You can also ask Dexter mid-conversation to promote the work into a Project — Dexter scopes the phases and opens the Project with your approval.

What’s Next

  • Model your operation: Your First Simulation builds a simulation model — the pipeline you just created can feed it directly
  • Plan production: head to the sidebar’s Planning surface if you have demand orders, inventory, and BOMs to work with
  • Build a shareable report: ask Dexter in Chat to turn your insight into a PDF or PowerPoint deliverable
  • Teach Dexter your operation: the more it knows about how your factory actually runs — told in Chat or edited directly in the Knowledge Base — the sharper every subsequent piece of work becomes
The pattern that just worked — upload, talk, get an artifact, approve — is the pattern the whole platform is built around. Every capability in the sidebar is a variation of it.