Skip to content
Back to work

02Production Cost & Waste Control

CleverOp Yield

Bakeries price their products from recipes. The bench rarely follows the recipe exactly. CleverOp Yield measures the difference and puts a price on it.

Role
Product, design & engineering
Timeline
September 2026
Platform
Windows 10/11 · fully offline
Status
v1.1.0 · installer built
Languages
English · العربية
  • Flutter
  • Desktop
  • Production
  • Cost Control
  • Business Software
CleverOp Yield
CleverOp Yield owner dashboard: loss against recipes over 30 days, items needing attention and the biggest losses.

01

Problem

A few percent extra butter, a batch that came out short, product thrown away at closing — each loss is small, but together they erase margin, and nobody can see where the money went.

  • Ingredient over-use goes unnoticed on a busy bench
  • Low-yield batches hide inside daily volume
  • Waste is recorded on paper, if at all
  • Prices change, but product costs are calculated once and forgotten

02

Product

Operators record what each batch really used and produced. CleverOp Yield compares it with the recipe, costs the batch at the prices of its day, and raises the unexplained part of the loss for the owner to review.

  • Batches

    A weigh-and-enter flow for operators, fully keyboard driven.

  • Waste

    Recorded on a batch or on its own, valued at the right price.

  • Anomalies

    An inbox of losses to explain, each reviewed with a cause and a note.

  • Dashboard

    Loss against recipes, what needs attention and the biggest losses.

  • Reports

    Variance, waste and product cost for any period, exported to PDF.

  • Catalog

    Ingredients, prices, products, recipes and intermediates like laminated dough.

  • Users & audit

    Administrator and operator roles, with a full audit log.

  • Backup & demo

    Validated backup and restore, plus a demo bakery kept in its own file.

03

Core Workflow

From recipe to review, every batch follows the same path — and every number on the way can be traced back to who entered it and how.

  1. 01

    Recipe

    Ingredients, quantities and prices, including intermediate products.

  2. 02

    Start batch

    Pick the product and planned output; recipe quantities are scaled.

  3. 03

    Weigh

    Enter the measured quantity per line — or confirm the recipe. Must-weigh ingredients can only be measured.

  4. 04

    Output & waste

    Good output and waste recorded against the batch.

  5. 05

    Variance

    Every line compared with the recipe; yield compared with plan.

  6. 06

    Batch cost

    Costed at the prices of its production date — frozen and exact.

  7. 07

    Anomalies

    Over-use, low yield, unexplained loss and statistical outliers raised automatically.

  8. 08

    Review

    The owner records a cause and a note. Every decision is audited.

04

Architecture

A local-first monolith: no server, no account, no internet. Everything runs in-process on one Windows PC, with the database in the user’s own profile.

  1. Interface

    • Flutter · Material 3

      Custom design system · Riverpod · go_router

    • Charts & PDF

      Hand-drawn charts · pdf export with bundled Arabic font

  2. Domain

    • Pure-Dart services

      Costing, variance, anomaly engine, roles enforced in services

    • Audit

      Language-neutral audit events, rendered in the reader’s language

  3. Data

    • SQLite via Drift

      Versioned migrations in one transaction, with a copy made first

  4. Windows

    • Native APIs via FFI

      CNG password hashing · Save/Open dialogs · Documents folder

    • Installer

      Per-user Inno Setup installer, no admin rights needed

05

Engineering Challenges

  1. 01

    Exact money on a desktop

    Money is stored in integer micro-units and quantities in milli-units. Costs are computed exactly with BigInt and rounded once, at the very end.

  2. 02

    A confirmation is not a measurement

    A typed quantity is measured; taking the recipe quantity is recipe-confirmed and never counted as a measurement. Reports show how much of production was actually weighed.

  3. 03

    Anomalies worth trusting

    Loss is judged only on fully priced batches. Outliers need |z| ≥ 3 against up to 20 earlier batches of the same product. A reviewed anomaly is raised again only if it gets worse.

  4. 04

    Costs that stay true over time

    Each line freezes its cost basis at the production date. Later price changes only fill gaps — they never rewrite history.

  5. 05

    Operators never see money

    Authorization lives in the domain services, not the interface. Operators never receive prices, costs or yield percentages — not even in raw data.

06

Reliability & Security

Reliability

  • Every schema upgrade runs in one transaction, after a copy of the database is made.
  • Backups are written with VACUUM INTO and validated before they are kept.
  • Restore accepts only genuine, undamaged backups — integrity, foreign keys and an exact schema allow-list are checked.
  • The current data is always set aside before a restore, so it can be undone.
  • 200 automated tests plus 56 golden-screenshot UI tests in English and Arabic.

Security

  • Passwords hashed with PBKDF2-HMAC-SHA256 through Windows CNG.
  • A session lock that covers dialogs and menus; unlocking re-verifies the user.
  • At least one active administrator is always kept, enforced in a transaction.
  • A full audit log of prices, recipes, batches, corrections, reviews, backups and exports.
  • Stated openly: anyone sharing the same Windows account can open the database file.

07

Technology

Interface
Flutter 3DartMaterial 3Riverpodgo_router
Data
SQLiteDriftVersioned migrations
Platform
Windows FFIpdfInno Setup
Quality
flutter_testGolden screenshotsflutter analyze

08

Screenshots

Screenshots from the built-in demo bakery — names and figures are fictional.

  • CleverOp Yield
    Completed batch with each ingredient line against the recipe, measured and confirmed markers, and batch cost.
    Batch detail — every line against the recipe, measured vs. confirmed, with a critical over-use flagged.
  • CleverOp Yield
    Operator batch entry screen with measured quantities and keyboard shortcuts.
    Operator view — a weigh-and-enter flow with keyboard shortcuts. No prices, no costs.
  • CleverOp Yield
    Variance report with expected and actual batch cost, loss against recipe and a breakdown by product.
    Reports — variance, waste and product cost for any period, exported to PDF.
  • CleverOp Yield
    The owner dashboard in Arabic with a right-to-left layout.
    Arabic, right to left — for owners and teams who work in Arabic.

09

Lessons

  • “Say what’s missing.”

    When a price is missing, figures carry a mark and a note instead of looking complete. An honest “incomplete” beats a confident wrong number.

  • “Offline is a feature.”

    No account, no server, no subscription. For a small bakery, software that just works on the office PC is the point.

  • “Design for the bench.”

    Operators work with flour on their hands. Keyboard flow and large, forgiving inputs matter more than features.

10

Current State

v1.1.0 “Trusted Production Capture”

Version 1.1.0 was built in September 2026 with a per-user Windows installer, and has been installed, upgraded and uninstalled on a real PC. Next: owner review, then first pilot sites.

  • One PC per site and one currency per installation.
  • The installer is not code-signed yet, so Windows SmartScreen shows a warning.
Next projectCleverOp ClinicModern Clinic Management Platform

Start a Project

Need software built around your workflow?

I design and build business applications with the same care you see here.

+961 71 279 930dbouk_9@hotmail.com