What does the Production Analyst need to work, and where does our data go?

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • DataForge
    Junior Member
    • 11.08.2025
    • 15

    #1

    What does the Production Analyst need to work, and where does our data go?

    I saw the PRODUCTION ANALYST menu in the web version. The idea of asking "Which line stopped the most last night?" instead of building a report is attractive for our managers. When I open it, it seems to need some configuration first.
    Before I bring this to our IT department I need to answer a few questions: what has to be set up, does it need an internet connection, and what data leaves our network? Also, how reliable are the answers? Our managers will take the numbers at face value.
  • admin
    Yönetici
    • 26.02.2025
    • 23

    #2
    Good questions to ask before rolling it out. Here's what you need to know.
    What it needs
    The Production Analyst uses Anthropic's Claude models. Configuration is done in SETTINGS > AI (the AI Settings page):
    - Anthropic API key: your own key. It must start with sk-ant-. The key is stored on the server and is never sent back to the browser.
    - Model: you can choose between the listed Claude models, for example the more accurate option or a faster, more economical one.
    Alternatively, your IT team can provide the key through the ANTHROPIC_API_KEY environment variable on the server; the page will then show the key source.
    Internet and data
    The Proveris server needs outbound internet access to the Anthropic API. When a user asks a question, the question and the production data needed to answer it are sent to the model. Your IT department should review this against your data policy, and the API key is under your company's control and account.
    Reliability
    The analyst answers from the records in your Proveris database, and the page says so: "Answers are generated from database records; verify figures before critical decisions." In practice, the quality of the answers depends on the quality of the data. If many downtimes are still under "No Reason Assigned", a question about downtime causes will reflect that.
    Our suggestion: start with a small group of managers, compare a few answers with the regular reports for the first weeks, and then widen access.
    If you'd like to discuss it with us before deciding, you can reach us at https://greensoft.ai/get-in-touch/

    Comment

    Working...