Selected work

See the workflow, scope, and evidence.

These case studies explain the operating problem, what Glitch was responsible for, what was built, and the limits of the available evidence. Measured results are published only when the baseline, period, and source are confirmed.

Workflow dossiers

Clear boundaries around every example.

Outcome labels describe the available evidence carefully. A completed deployment does not automatically mean a result has been measured.

01

Connecting garment orders from costing to dispatch and payment

Evidence: Expected operational benefit

A garment manufacturing business needed a clearer way to connect order information, costing, procurement, pre-production, production, dispatch, GRN, payment, alerts, and management reporting across separate files, teams, and owners.

Business problem

  • Repeated follow-up for latest status
  • Costing approval and procurement readiness were disconnected
  • Production, dispatch, GRN, and payment were maintained separately
  • Management lacked one reliable order-level view

Previous process

  • Order information in one tracker
  • Costing and BCS maintained separately
  • Procurement used another structure
  • Dispatch, GRN, payment, and reporting required follow-ups

What Glitch built

  • Connected tracker around one Internal Order ID
  • Order, costing, approval, procurement, readiness, production, and dispatch tracking
  • Role-based editing, protected fields, standard dropdowns, alerts, and dashboard

Key controls

  • Role-based access
  • Protected formula fields
  • Required approvals
  • Shared master data and standard status definitions
  • Validation rules and delivery-risk alerts

Expected operational benefit

  • Less manual consolidation
  • Clearer ownership
  • More consistent updates
  • Earlier visibility into procurement and delivery risk

Possible measurement method

  • Weekly reporting hours
  • Repeated update requests
  • Required-field update rate
  • Pending approval age
  • Delivery risks identified before threshold

Limitations

The system depends on timely updates, correct order identifiers, stable field definitions, clear ownership, and user adoption. It should not be presented as a complete apparel ERP unless the implemented scope supports that claim.

Discuss a connected operations workflow
02

Moving a low-stock exception into procurement action

Evidence: Expected operational benefit

A manufacturing workflow required a clearer connection between inventory data, low-stock detection, approval, supplier communication, purchase-order generation, invoice handling, stock updates, and management visibility.

Business problem

  • Low stock identified too late
  • Quantities and suppliers required manual approval
  • RFQ and purchase-order preparation was repeated
  • Management lacked a procurement-exception view

Previous process

  • Inventory checked manually
  • Low-stock information shared in messages or email
  • Approvals and supplier communication handled separately
  • Invoice and stock updates handled independently

What Glitch built

  • Reorder-level checks and low-stock detection
  • Approval request with supplier and quantity selection
  • Supplier-grouped RFQs, purchase orders, invoice review, stock update, and dashboard activity

Key controls

  • Human quantity and supplier approval
  • Required-field validation
  • Email and document logs
  • Review before irreversible stock updates
  • Approval-decision record

Expected operational benefit

  • Faster movement from exception to action
  • Less repeated handling
  • Clearer approval history
  • Better visibility into low stock and purchase activity

Possible measurement method

  • Time to approved action
  • Manual handoffs per request
  • RFQ preparation time
  • Missing-field completion rate
  • Invoice extraction correction rate

Limitations

The workflow depends on reliable stock data, consistent item identifiers, accurate supplier information, approved commercial rules, and stable document formats. Invoice extraction should retain human review.

Discuss an inventory or procurement workflow
03

Giving technical document reviewers a structured first pass

Evidence: Product capability

Technical documents and scanned files can contain spelling, grammar, terminology, and OCR-related issues. Reviewers may spend significant time on basic checks before technical content.

Business problem

  • Senior reviewers repeat similar checks
  • Generic tools flag correct technical terms
  • Scanned files are harder to review
  • Review history and accepted or ignored issues have limited visibility

Previous process

  • Manual document review
  • Additional effort for scanned files
  • Technical terms repeatedly flagged
  • No reusable terminology list

What Glitch built

  • PDF extraction and OCR
  • Possible issue highlighting
  • Technical terminology whitelist
  • On-page Accept, Ignore, and Add to approved terms actions
  • Review history and user-controlled final decision

Key controls

  • Human reviewer remains in control
  • Approved terminology handling
  • Recorded review actions
  • Final technical, legal, quality, and client judgement remains human

Expected operational benefit

  • More structured first pass
  • Faster identification of possible issues
  • More consistent terminology handling
  • Reduced repetitive checking

Possible measurement method

  • OCR success by document type
  • Reviewer acceptance rate
  • False-positive rate
  • Processing time
  • Accuracy on approved test set

Limitations

Results depend on scan quality, page layout, language, terminology, document complexity, and provider behaviour. The system cannot guarantee every issue will be detected or replace professional review.

Discuss a document review workflow
04

Structuring work across executor, checker, and approver roles

Evidence: Expected operational benefit

Some engineering and project workflows require work to move through execution, checking, approval, rejection, and rework, while management needs visibility into overdue work, workload, and quality.

Business problem

  • Task assignment and review stages were inconsistent
  • Overdue work was difficult to identify early
  • Rework history was fragmented
  • Management lacked a structured workload view

Previous process

  • Tasks assigned manually
  • Execution and review handled through separate communication
  • Approval history difficult to track
  • Managers depended on follow-ups for current status

What Glitch built

  • Admin, executor, checker, and approver roles
  • Defined status stages, rejection, and rework paths
  • Overdue visibility, pause tracking, workload views, activity records, and performance indicators

Key controls

  • Role-based permissions
  • Sequential review stages
  • Required approval flow
  • Rework instructions and status history
  • Controlled data access and human performance review

Expected operational benefit

  • Clearer responsibility
  • More consistent review stages
  • Better visibility into overdue work
  • Stronger record of rejection and rework

Possible measurement method

  • On-time completion rate
  • Time in workflow state
  • Rework and rejection rate
  • Workload distribution
  • Overdue ageing

Limitations

Performance indicators are only as reliable as task definitions, user updates, and context. They should not become automatic employee judgement without management review and appropriate policy.

Discuss a role-based workflow

Trust and confidentiality

Similar work is useful evidence, not a promise of the same outcome.

Every organisation has different users, systems, data quality, controls, and adoption conditions. Discovery is required before Glitch can confirm scope, timeline, or expected benefit for a new project.

Screenshots, client names, documents, and operational data are published only with permission. Where confidentiality is required, Glitch may use redacted interfaces, anonymised workflow descriptions, and limited context.

Discuss a similar workflow

Which project looks closest to your workflow?

Share the current process, people involved, systems in use, and where work is slowing down. Glitch will first assess whether the project is similar enough to use as a practical reference.