Establishing Orbit
Capabilities
All capabilities Operations & Transformation Consulting Shop Floor Digitisation Machine Connectivity & Integration Custom Software, Data & AI Adoption, Support & Scale
Company
The Connected Floor The Orbit Method Industries About Book an Audit →
C04 — Custom Software, Data & AI

For everything that doesn't come in a box.

Packaged platforms cover the common eighty percent of a factory well. The remaining twenty is where your competitive difference usually lives — the scheduling logic nobody else has, the customer portal your key account is asking for, the model that predicts your specific failure mode. That's what we build.

The straight answer

When should a manufacturer build software rather than buy it?

Build when the process is genuinely differentiating, when no product covers it without heavy compromise, when integration into an existing estate is the harder problem, or when licence economics at your scale make ownership cheaper. Buy when the need is standard and well served. Most plants end up with a hybrid: a configured platform for common functions plus bespoke software for the parts that are specific to them. OrbitX delivers both and has no commercial reason to steer you toward either.

When this is relevant

You probably need this if…

  • A product almost fits, but the workarounds are starting to outnumber the features you use.
  • Your planning or scheduling logic lives in one very clever spreadsheet and one very key person.
  • Customers or suppliers are asking for portal access you can't provide.
  • You have data from several systems and no single place that reconciles it.
  • You want predictive maintenance and don't know whether your data supports it.
  • You're being quoted per-seat licence costs that don't work at your headcount.
What we fix

Problems we address

  • Off-the-shelf software bent so far out of shape that upgrades become dangerous.
  • Critical business logic living in spreadsheets with no version control and one owner.
  • Reporting built separately in each system, so no two numbers agree.
  • AI projects started before the underlying data was reliable enough to model.
  • Bespoke systems built by a departed contractor with no documentation.
  • Integrations written point-to-point until nobody can safely change anything.
Our approach

How we work through it.

  1. Establish the data before the model

    Predictive work needs history, labels and reliable capture. Where those don't exist yet we say so and sequence the instrumentation first, rather than selling a model that will produce confident nonsense.

  2. Build the smallest thing that proves the value

    A narrow first release in production beats a comprehensive one in development. We ship something people use within weeks, then extend against real feedback.

  3. Design for handover from day one

    Documentation, tests, readable architecture and a stack your team can hire for. You should be able to take the codebase to another firm without archaeology — that's a deliberate constraint, not an accident.

  4. Integrate through contracts, not shortcuts

    Every interface is defined and documented so systems can be replaced independently later. The shortcut version is faster this quarter and expensive for the next five years.

  5. Keep your data yours

    Your operational data stays in your environment, in open formats, with export you control. No hostage-taking through proprietary storage.

Scope of work

What sits inside C04.

C04.01

Bespoke Operational Applications

Software built around a process that no product covers properly.

  • Requirement and process discovery
  • Application design and build
  • Role-based access and approvals
  • Offline-capable mobile applications
  • Deployment and hypercare
C04.02

Operator & Supervisor Interfaces

Screens designed for gloved hands, poor lighting and a shift running behind.

  • Shop-floor terminal interfaces
  • Tablet and handheld applications
  • Barcode and RFID capture
  • Large-format line displays
  • Accessibility and language support
C04.03

Planning & Scheduling Tools

Getting the logic out of the spreadsheet and out of one person's head.

  • Production scheduling and sequencing
  • Capacity and constraint modelling
  • Changeover optimisation
  • Material requirement alignment
  • Scenario comparison
C04.04

Unified Data Platform

One reconciled operational record underneath everything else.

  • Data architecture and modelling
  • Pipelines from machines and systems
  • Historisation and retention design
  • Data quality monitoring
  • Governance and access control
C04.05

Analytics & Reporting

Numbers people actually act on, at the level they act at.

  • Operational and executive dashboards
  • Automated report generation
  • Multi-plant benchmarking
  • Cost and margin analytics
  • Self-service reporting layer
C04.06

Applied AI & Prediction

Models built where the data genuinely supports them, and not where it doesn't.

  • Predictive maintenance models
  • Quality and defect prediction
  • Anomaly detection on process parameters
  • Demand and capacity forecasting
  • Natural-language querying of operational data
C04.07

Portals & External Systems

Extending the operating record beyond your four walls.

  • Customer order and quality portals
  • Supplier collaboration portals
  • Field service and warranty systems
  • Dealer and distributor interfaces
  • Regulatory submission support
C04.08

Product Engineering

For manufacturers whose product itself is becoming connected.

  • Connected product architecture
  • Device firmware integration
  • Telemetry and fleet management
  • Companion applications
  • Software licensing and update paths
C04.09

Modernisation & Rescue

Taking over software that was built badly or abandoned.

  • Legacy application assessment
  • Incremental modernisation
  • Codebase documentation and recovery
  • Re-platforming
  • Takeover of unsupported systems
Deliverables

What you get

  • Working software in production, deployed in phases rather than one release.
  • Documented architecture, data model and interface contracts.
  • Source code and infrastructure you own outright.
  • Automated tests and a deployment pipeline.
  • A data platform reconciling operational data across sources.
  • Model documentation covering assumptions, accuracy and known limits, where models are in scope.
  • Handover materials and training for your internal team.
Outcomes

What changes

  • Business logic moves out of spreadsheets and single-person dependency.
  • Reports agree with each other, because they read from one reconciled record.
  • Customers and suppliers self-serve instead of emailing your team for status.
  • Failures get flagged before they stop the line, where the data supports prediction.
  • You can change or replace one system without breaking six others.
  • Your team can maintain and extend what we built without us.
Engagement

How we partner on this.

ModelScopeTypical duration
Build vs. Buy AssessmentIndependent analysis of whether to build, buy, extend or wait — with costs modelled across five years.1–2 weeks
Discovery & PrototypeProcess discovery, solution design and a working prototype to validate the approach before committing to a build.3–5 weeks
Build ProgrammeFull design, build, integration and deployment, delivered in phased releases.3–9 months
Data Platform ProgrammeData architecture, pipelines, quality monitoring and reporting layer.2–6 months
Modernisation EngagementAssessment and incremental replacement of a legacy or abandoned system.Variable
Questions

Frequently asked

Often you shouldn't, and we'll say so. Building is right when the process is genuinely differentiating or when no product fits without damaging compromise. For standard functions, configuring a proven platform is faster, cheaper and lower-risk. The build-vs-buy assessment exists to answer this honestly — and it recommends 'buy' more often than it recommends 'build'.
You do, outright, including infrastructure definitions and documentation. We don't retain licensing rights over software you paid us to build, and we don't hold your data in a proprietary format. You should be able to move to another firm without a negotiation.
Maybe. Prediction requires failure history, labelled events and reliable condition data — and most plants starting out have none of the three. The honest sequence is usually: instrument the assets, capture clean history for six to twelve months, then model. Anyone promising predictive maintenance in week one on an uninstrumented plant is selling you a dashboard with an ambitious name.
We choose based on what your team can realistically maintain, integration requirements and deployment constraints, rather than on what we most enjoy writing. Where you have an existing in-house capability, aligning to it is usually worth more than any technical advantage of the alternative.
You have the code, the documentation, the tests and the infrastructure. We'll run a structured handover to your team or to another firm. This is designed in from the start — a services firm whose retention depends on you being unable to leave is a bad partner, and we'd rather compete on being worth keeping.
Related capabilities
Start Here

Start with the audit, not the software.

A fixed-scope engagement. We walk your floor, map where data breaks down, and hand you a costed roadmap — whether or not you build any of it with us.