Software demos show the best day. This is the ordinary one: what you set up, what the platform does while you're not looking, and what you're left holding when someone asks for proof.
The first two steps are the implementation. The last four are what the platform does from then on — most of it without anyone opening a spreadsheet.
Organisations, properties, departments and the asset register — the backbone everything else attaches to.
Compliance items and PPM tasks with their frequencies, service types and checklists.
Due status, calendar and planner show what's coming and what's slipped, filtered to your sites.
Certificates and engineer reports land against the record — and go into the scan queue.
Remedial actions turn into owned, prioritised work with a link back to the source finding.
Dashboards, job reports and Excel exports — evidence produced in minutes, scoped to who's asking.
An engineer's report arrives as a PDF. Historically that meant someone reading it, deciding what mattered, and typing the important bits into a system — slowly, inconsistently, and often not at all.
In PriComp the upload queues the document. A background worker reads it and produces a structured record: who attended, which site, what was tested, what was found, what needs doing, the outcome, the next due date. If the report indicates work is needed, an action required flag appears on the compliance item — so the risk surfaces without anyone opening the attachment. Remedial actions can be turned into linked jobs in one click, pre-filled from the finding. Assets mentioned in the report are checked against the register and can be created from it, categorised sensibly.
Then a human checks it. That order matters: the model drafts, a person confirms. Nothing lands in the compliance register unreviewed.
Most estates have years of certificates sitting in folders. The queue is designed to work through them unattended: it paces itself around third-party rate limits, retries what's transient, gives up cleanly on what isn't, and re-checks itself every five minutes so a mid-run deploy can't strand anything. A backfill can be narrowed to one document type at a time, because every scan is a real cost.
A compliance record has to still be there, still be right and still be readable in five years. So the data layer is deliberately conventional and the delivery process is deliberately strict — and the newer technology goes where it genuinely removes work, which today means reading documents.
A browser-based platform — nothing to install, nothing to keep patched on your machines.
Google's framework for large, structured business applications. It is why the platform can carry a dozen modules without turning into a maze.
JavaScript with a safety net, used on both sides of the platform. Types catch mistakes at build time rather than in front of a compliance officer.
The API behind every screen. One documented interface serves the web app, the exports and anything you integrate later.
A relational database, because compliance data is relational: a site has assets, an asset has services, a service has a certificate.
Registers and dashboards download as real spreadsheets, generated server-side and scoped by exactly the same access rules as the screens.
Portfolio health at a glance, with a yearly tracker built from stored monthly snapshots rather than reconstructed from current dates.
The headline capability — and the part with the most safeguards around it.
Safety data sheets and engineer reports read into structured records: provider, visit, assets tested, findings, remedial actions, outcome and next-due dates.
Processing runs in the background, paces itself around rate limits, retries transient failures and sweeps for stranded documents every five minutes.
Queued → Scanning → Done, with a counter in the top bar and a super-admin control to process the existing back-catalogue.
Extraction fills the form; a person checks and saves it. Nothing enters your register on the say-so of a model alone.
The controls that make a shared system safe to share.
Super-admin, system user and engineer roles, each with the permissions their job actually needs — and no more.
People see the sites they are responsible for. The filter is applied on the server, so exports and reports can never reveal more than the screens do.
Records are retired, never destroyed. The audit trail of who recorded what, and when, survives deletion — which is the whole point of an audit trail.
Self-service password reset and first-login tracking, so you can see who has actually started using the system rather than who was invited.
Organisation → property → department, so a group with twelve sites and a single site with four departments both fit the same model.
A bug-reporting tool built into the platform, so problems come back from the field with context instead of dying in a corridor conversation.
How changes reach you — and how we know they are safe before they do.
Every change built, tested and deployed automatically, with a visible application version — so support conversations start from facts.
Curated suites on both the front end and the API run on every pull request. A feature is not finished until its test exists and passes.
Every schema change is recorded as a dated migration and applied once. Nothing changes the shape of your data without a written record of it.
Database backups are part of the operating routine, not an afterthought — because a compliance record you cannot restore is not a record.
Score your position first, or skip straight to a walkthrough against your own sites, asset types and statutory regime.