What is implemented
A Streamlit page loads data, calculates suggested quantities, shows a replenishment table, lets Tim edit final quantities, exports an order CSV, and stores open-order notes.
It is trying to turn stock, sales, supplier files, and ordering rules into a short list of products Tim can review and order. It is not yet a full inventory system, customer relationship manager, or automatic purchasing system.
The current page is useful but overloaded because it mixes the working tool, technical controls, and future product ideas. This is the clean distinction.
A Streamlit page loads data, calculates suggested quantities, shows a replenishment table, lets Tim edit final quantities, exports an order CSV, and stores open-order notes.
Open the app, check the data, review the short list, change anything that looks wrong, send the order, and tell us what was confusing or inaccurate.
Start with “Your next order,” hide technical detail, show only decisions that need attention, and make inbound orders part of the same simple loop.
These are the current functions, translated into the job they perform for Tim.
Use the data-center API for the MYOB item master, or upload an item export as an override or fallback.
Bring in on-hand, on-order, committed, and available quantities from MYOB stock data.
Use API sales data, a local cache, or uploaded sales files to estimate how quickly each product sells.
Choose lead time, safety stock, and the sales period used to estimate demand.
Compare target stock with current coverage and apply MYOB reorder quantities to produce a suggested quantity.
Call out negative available stock, suspicious inputs, order mismatches, and items that cannot be matched safely.
Filter the replenishment table, inspect a product, and change the final quantity before sending anything.
Download a supplier-ready CSV, optionally save it as an open order, then update received quantity, ETA, status, and notes.
The page is one long workbench. The left sidebar contains setup and data controls. The main area contains the decision table and several technical or follow-up sections below it.
This is a structural map, not a screenshot. The exact counts and rows change with Tim’s data.
The current pilot is not asking Tim to become a system administrator. It is asking him to test whether the proposed order is useful enough to act on.
The API should load items, stock, and sales. Uploads remain available when a source needs to be overridden.
Look for source errors, missing sales, stale cache information, or an unexpected stock picture.
Use Need Reorder, the flagged inspector, and table filters to focus on products that need a decision.
Check the product, available stock, incoming stock, demand, and suggested quantity. Edit Final qty when needed.
Download the supplier CSV. Save it to the tracker if it represents an order Tim has sent.
Tell us what was right, wrong, missing, slow, or harder than the old process. That feedback shapes the next version.
The expected behavior is not “trust the number.”
It is “use the number as a starting proposal, then tell us whether the facts and the workflow match how Derwent actually buys.”
| What is on the page | What it is for | What Tim should do in the pilot |
|---|---|---|
| Planning inputs in the sidebar | Control lead time, safety stock, and demand assumptions. | Accept the saved defaults unless a real Derwent policy needs changing. Do not treat every control as a required setup step. |
| API status, cache details, and uploads | Make the data source visible and provide a fallback. | Confirm that the data is usable. Upload only when the source needs an override or the API is incomplete. |
| KPI row and flagged warning | Summarize the catalogue and surface possible data problems. | Use “Need reorder” and flags as the entry point. The totals are context, not the work itself. |
| Replenishment table | Put one decision row per product in front of Tim. | Review the short list, inspect why, edit Final qty, and select lines for export. |
| Reconciliation, planner, formulas, diagnostics | Support follow-up, explanation, and troubleshooting. | Use these when a decision needs investigation. They are not supposed to be read top-to-bottom every time. |
The redesign should keep the current calculation and data integration while changing the first impression and the order of attention.
Open on “Your next order” or “Nothing urgent,” not on a catalogue of controls.
Keep settings, uploads, formulas, mapping, and technical logs available but out of the main path.
Make review, export, sent orders, incoming stock, and reconciliation feel like one continuous task.
Bottom line: the current app is a working pilot instrument. Its job is to generate real buying decisions and learn from Tim’s corrections. The next UI should make that loop obvious without changing the underlying goal.