Case studies
Case studies of custom software built for real operations.
Real systems delivered across medical, engineering, finance, logistics, project management and education businesses — built on Claris FileMaker, Supabase and REST integrations. Every system here went live, carried real operational load, and is either still running or still the pattern we start from.
Browse by sector. Each card sets out what the operation looked like before, what the system does, and what changed afterwards. The detail behind each one — data model, integrations and reporting — is what we walk through on a discovery call.
- 20+
- Years building
- 100+
- Systems delivered
- 6
- Industries served
- 12
- Systems shown here
Practice & patient administration suite
Scheduling, clinical records, invoicing and reporting for multi-practitioner practices, with role-based access and a full change history behind every record.
The starting point
Practices ran appointments on paper diaries, clinical notes in a shared folder and billing in a separate accounting package. Nothing spoke to anything else, so every invoice had to be assembled by hand and nobody could say what a practitioner had actually done that month.
What the system does
- Patient, practitioner and appointment records with conflict and double-booking checks
- Clinical notes and attachments held against the patient, with per-practitioner visibility
- Visit-to-invoice generation driven by practitioner-specific fee schedules
- Practice dashboards for appointments, revenue and outstanding balances
What changed
Reception books and bills from one screen. Month-end no longer depends on someone rebuilding the numbers from diary pages, and every clinical and billing change carries a name and a timestamp.
Typical modules
Built on
Laboratory workflow tracker
Specimen intake through to result release, with barcode handling, turnaround-time monitoring and instrument integration over REST.
The starting point
Specimens arrived by courier and were logged on a printout. When a client asked where a sample was, someone walked the bench to find out, and turnaround figures were estimated rather than measured.
What the system does
- Barcode intake with label printing and chain of custody at every hand-off
- Result capture pulled directly from instruments over REST
- Turnaround clocks per test type with escalation when a target is at risk
- Release and reporting with the referring clinician's details attached
What changed
Any sample can be located in seconds, turnaround is measured rather than remembered, and results reach referring practices without being re-typed.
Typical modules
Built on
Job costing & production control
Quote to job card to invoice, with labour and material capture, WIP valuation and margin analysis per job.
The starting point
Quotes lived in word processor files, job cards on the floor and costs in the accountant's spreadsheet. By the time a job was costed it was finished, so the margin was known only after the money had gone.
What the system does
- Quoting with costing sheets that convert straight into a job card
- Labour and material capture at the bench, on the floor or on site
- Work-in-progress valuation that updates as costs land
- Margin reporting per job, per customer and per product family
What changed
Jobs are costed while they are still running. Unprofitable work is spotted early, quoting improves against real history, and month-end reconciliation shrinks to a review.
Typical modules
Built on
Asset & maintenance register
Planned and reactive maintenance, service history, spares consumption and compliance certificates in a single register.
The starting point
Maintenance was reactive and documented in whatever folder the technician happened to use. Compliance certificates surfaced the week before an audit, and spares use stayed invisible until a stock-out stopped a line.
What the system does
- Asset register with hierarchy, location and complete service history
- Planned maintenance schedules that raise work orders and reminders automatically
- Reactive fault capture from a mobile device on the floor
- Spares consumption linked to assets, with compliance certificates stored on the record
What changed
Unplanned downtime becomes visible and forecastable, audits pull evidence straight from the asset record, and spares are ordered against actual consumption.
Typical modules
Built on
Client onboarding & compliance desk
Structured onboarding, document checklists, expiry monitoring and audit-ready record keeping for regulated client relationships.
The starting point
Onboarding a new client meant email chains, a shared drive and a checklist nobody could see. Expiring documents were discovered by chance, and preparing for an audit meant reconstructing months of correspondence.
What the system does
- Structured onboarding workflow with role-based tasks and escalation
- Document checklists with expiry monitoring and advance notification
- Template-driven document generation with version history
- Compliance reporting across the whole client book
What changed
No client is activated with a document missing, expiries are caught weeks ahead, and an audit pack becomes a report rather than a project.
Typical modules
Built on
Commission & reconciliation engine
Rule-driven commission calculation, statement generation and reconciliation against imported bank and provider files.
The starting point
Commission rules lived in the heads of two people and a set of spreadsheets. Statements took days to produce, disputes took longer, and reconciliation against provider files was manual line matching.
What the system does
- Commission rule tables per provider, product and effective date
- Scheduled calculation runs that produce statements and journals
- Automated import and matching of bank and provider files
- Dispute and adjustment history held against each policy
What changed
Statement runs drop from days to hours, disputes are answered from the record, and reconciliation leaves a short exception list instead of a stack of printouts.
Typical modules
Built on
Dispatch & consignment tracking
Load planning, driver allocation, live consignment status and proof of delivery captured in the field.
The starting point
Loads were planned on a whiteboard, drivers carried paperwork, and the office answered 'where is my consignment?' by phoning the driver. Proof of delivery came back in a bag of dockets.
What the system does
- Load planning with vehicle and driver allocation
- Consignment status updated from the field on a mobile device
- Proof of delivery captured at the door — signature, photo and exceptions
- Client-facing consignment status exposed over REST
What changed
The office stops phoning drivers, proof of delivery is in the system the moment it is captured, and billing follows what was actually delivered.
Typical modules
Built on
Warehouse & inventory control
Bin-level stock, goods received, picking, stock takes and variance reporting across multiple sites.
The starting point
Stock sat in bins across several sites with no reliable picture of where anything was. Picking depended on whoever had been there longest, and a stock take stopped the warehouse for days.
What the system does
- Bin and location control across multiple sites
- Goods received, picking, packing and dispatch captured on a scanner
- Cycle counting with variance reporting and approval
- Reorder points and inter-site transfer control
What changed
Pickers find items without asking, purchasing trusts the stock figure, and counting happens continuously instead of shutting the warehouse down.
Typical modules
Built on
Project management & resource planning
Resource planning, task tracking, milestone control and budget oversight for project-based teams running several engagements in parallel.
The starting point
Several engagements ran in parallel with resource conflicts discovered at the morning stand-up, budgets held in a separate spreadsheet and status rebuilt by hand for every steering meeting.
What the system does
- Engagement setup with phases, tasks and dependencies
- Resource planning across concurrent projects with over-allocation flags
- Budget, commitment and actual with variance per engagement
- Risk register and status reporting prepared for steering meetings
What changed
Resourcing conflicts surface before they cost money, budgets are tracked against actuals rather than guesses, and status reporting is generated instead of assembled.
Typical modules
Built on
Student & course administration
Student and course administration, assessment tracking, attendance and certification management for training providers.
The starting point
Enrolments, attendance and assessment results lived in separate files. Certification reporting took most of a term to compile, and nobody could quickly say who had completed what.
What the system does
- Student, cohort and course records with full enrolment history
- Attendance registers and assessment tracking per module
- Prerequisite and progression checks before a student advances
- Certificate generation with verification records behind each one
What changed
Enrolment through to certification runs on one record, attendance and assessment gaps are flagged while a course is still running, and certification reporting is available on demand.
Typical modules
Built on
Integration & API middleware layer
A reusable connector framework linking FileMaker, Supabase, accounting packages and third-party REST services — with conflict handling and audit logs.
The starting point
Every new connection between systems was written as a one-off script. When one system changed, the break was discovered because someone noticed a missing invoice, and there was no record of what had moved where.
What the system does
- Reusable connector framework for accounting, couriers, payment gateways and internal services
- Scheduled and event-driven sync with retry and conflict handling
- Configuration per client rather than hard-coded endpoints and credentials
- Audit log and monitoring on every cross-system movement
What changed
New integrations become configuration rather than new code, failures are caught by monitoring instead of by users, and every movement between systems is traceable.
Typical modules
Built on
Semantic search & AI assistant layer
Retrieval-augmented search and assistants over quotes, job cards, notes and correspondence — grounded in each client's own records and permissions.
The starting point
Twenty years of quotes, job cards, notes and correspondence already held the answers, but finding them meant knowing the exact wording the original author used, or asking a colleague who might have retired.
What the system does
- Embeddings generated and stored in Postgres with pgvector as records change
- Retrieval-augmented answers assembled from the client's own records
- Citations that link every answer back to its source record in FileMaker
- Permission-aware querying, so users only ever see what they already could
What changed
Staff search by meaning instead of keyword, answers arrive with the source record attached, and the assistant honours the same permissions as the system behind it.
Typical modules
Built on
Common ground
The problems we are brought most often.
Across twenty years and six industries, the same handful of situations keep appearing. They are rarely dramatic — usually a process that works, but only because a specific person is holding it together. That is the point at which a custom system earns its keep.
The spreadsheet that runs the business
One workbook holds pricing, another holds stock, a third holds the customer list, and the person who maintains them is the only one who fully understands the formulas. It works until they are on leave, until the file is copied, until the business outgrows it — which is exactly where most of our projects begin.
Systems that never speak to each other
Accounting says one thing, the operational system says another, and the difference is reconciled by someone with time. We connect what you already own over REST, scheduled syncs or a middleware layer so the same number appears everywhere without being re-typed.
No way to prove what happened
When a record changes, nobody can say who changed it, when, or what it said before. For regulated businesses that is a compliance problem; for everyone else it is an argument you cannot settle. We build the audit trail in from the start rather than bolting it on.
Reporting that takes days
The figures exist, but producing them means exporting, pasting and re-keying until someone is free. When the report is a button rather than a task, management decisions start happening weekly instead of monthly.
Process held in people's heads
The way work actually gets done was never written down, so it cannot be trained, checked or improved. Our requirement sessions surface it, and the system becomes the written-down version of how your business runs.
Growth that breaks the current tool
A system built for five users and one site rarely survives fifty users and three sites. Sometimes that means extending what you have; sometimes it means moving the heavy lifting to Postgres and Supabase while keeping the interface your team already knows.
Every build
What every system ships with.
These are not extras quoted at the end of a project. They are the baseline of how we build, because a system that cannot be audited, backed up, restricted or supported becomes a liability whatever else it does well.
Where a client needs more — a public portal, vector search, a mobile app carried by field teams — we add it on top of the same operational data rather than starting a second system alongside the first.
Role-based access
People see the records and fields their job requires. A receptionist, a practitioner and a practice manager open the same system and get three different screens.
Full audit trail
Who changed what, when, and the value before the change — stored with the record, not in a log file nobody reads.
Validation at entry
The expensive errors are the ones prevented at the keyboard: required fields, controlled lists, date and quantity rules, and duplicate checks that stop bad data reaching a report.
Reporting built for decisions
Dashboards and exports designed around the question your team actually asks, not a generic report pack.
Backup and tested restores
Offsite daily backups, with restores tested on a schedule. A backup nobody has ever restored is a hope, not a plan.
Hosting and licences
We source and administer FileMaker licences, deploy FileMaker Server on your infrastructure or ours, and monitor it.
Documentation and training
Written handover for administrators, on-screen guidance for users, and training for the people who do the work every day.
Support under an SLA
Response times, coverage hours and a maintenance agreement, so the system keeps working after go-live rather than drifting.
How a project runs
From first call to a system you own.
The order matters. Requirement definition before code, a working slice before the full build, and your sign-off before the next phase. It keeps cost tied to something you have actually seen.
- 01
Discovery call
Thirty minutes on the process that is causing the pain. We establish scope, appetite and whether the problem is worth building for — and say so if it is not.
- 02
Requirement definition
Workshops with the people who actually do the work. We map how it flows today, where it breaks, and what the system must do — written down, prioritised and signed off before any code is written.
- 03
Phased build with demos
We build the highest-value slice first and show you working software at regular intervals. Direction is corrected while it is still cheap to correct.
- 04
Testing and training
Your own people test against their own real records, with parallel running where the process cannot afford a hard cut-over. Training happens close to go-live so it sticks.
- 05
Go-live and support
Deployment, data migration and a supervised first week. Then an SLA, planned enhancements and platform updates as your operation changes.
The full consultation method — what each session asks for, what it produces and why it matters — is set out on the consultations page.
See the consultation process
By industry
Six industries, and what each one actually needs.
A system for a practice and a system for a workshop share a data model shape but very little else. Knowing which questions to ask in the first session is what two decades of delivered work buys you.
Medical
Practices and laboratories are regulated, time-sensitive and document-heavy. The work centres on records you can trust: who saw the patient, what was recorded, what changed and when — with access limited to the people entitled to see it.
Engineering
Job shops and project houses live or die on whether a job was quoted correctly and whether the margin survived production. We connect quoting, job cards, materials and costing into one chain so the number is visible while the job is still on the floor.
Finance
Onboarding, compliance and commission are process work that punishes shortcuts. Our systems enforce the checklist, watch the expiry dates, calculate the rules and produce evidence an auditor will accept without a phone call.
Logistics
Freight and warehousing are decided in the field, so the system has to go there. Barcode capture, mobile status updates and proof of delivery at the door turn operations into data you can bill from and report on.
Project Management
Running several engagements at once is a resourcing and budget problem before it is a task problem. We track allocation, commitments and variance so conflicts surface while there is still time to move someone.
Education
Training providers carry enrolment, attendance, assessment and certification obligations. One student record that runs from application to certificate removes most of the administration — and most of the reporting panic.
Custom or packaged
How custom development compares with buying a package.
We are honest about this in discovery, because the wrong answer costs a client more than a lost project. Sometimes a packaged product is the right call. Often the sensible answer is both — the statutory part packaged, the operational part built around how you actually work.
Whichever route fits, the integration work is usually the largest hidden cost. We scope it during selection rather than after, so the figure you approve includes connecting the new system to what you already run.
FAQs
Questions about projects like this
How we run a project, what you own at the end of it, and how the systems above were delivered.
Your system next
Most of these started with a spreadsheet and a frustrated team.
Bring us the process that eats your week. A 30-minute discovery call is usually enough to see which of these patterns fits.