Loading...
Loading...
Fluids capability
Most of what goes wrong in a plant is not a mistake. It is a hand-off nobody picked up. Fluids gives every one of 65 stages an owner, a deadline and a named person it escalates to — and none of that is written in code, so you change it yourself on a Tuesday afternoon.
A cylinder comes back from the filler and sits in inward for two days because the person who normally logs it was on leave. A requisition waits for an approval that the approver never saw. A quote goes quiet and nobody notices until the customer buys elsewhere. None of these is a failure of software or of care — they are the ordinary cost of work that is handed from one desk to another with nothing counting the time in between. That counting is what this module does, and it is the reason a plant on Fluids stops discovering delays at the point where a customer complains.
How it works
Configured today
These are shipped defaults, not illustrations. Every hour figure, every role and every escalation target below is editable from an admin screen without a release.
Also notified — QC, Branch Manager, Admin
A cylinder holding out-of-spec gas is the one thing that must not sit. It raises the rework job and tells four desks at once.
A customer's order is waiting on somebody choosing which bottles will fill it.
The one hand-off whose escalation target is not the branch manager — because the branch manager is the one holding it up.
Also notified — Inventory
Gas has come back from a third party. Nothing ships until it has been tested.
An exception is sitting in one person's queue. Twelve hours, then it goes over their head.
The cylinder is ready to leave. From here the only thing between it and the customer is a vehicle.
Ten working days to follow up a quote, twenty before the manager hears about it. A very different clock, on exactly the same engine.
The validity date has passed. Extend it, re-quote it or mark it lost — but do not let it rot silently.
Design decisions
Nothing in the product writes a notification by hand. A step records that an item reached a stage, and the database decides — from your rule table — who hears about it and by when. That is why changing a rule takes effect on the very next event and never needs a release.
Due time is the moment the thing happened plus the hours you configured. A task raised at six on Friday evening is already eleven hours old on Saturday morning, and the screen says so — rather than starting the count when somebody finally looks.
Past the escalation window the original is marked escalated and a fresh copy is addressed to the role you nominated, linked back to the one it came from. Most escalate to the branch manager. Requisition approval escalates past the branch manager to admin, and a manager review escalates to super admin.
If nobody at that branch holds the role a rule names, the notification is re-addressed to the branch manager instead of being raised against nobody. Every substitution is recorded — so the report also tells you which roles you are short of.
A task belongs to a role, so whoever is on shift can pick it up. Your own read state is tracked separately from whether the task has actually been actioned, so marking it read does not make it look done to everyone else.
Every alert that could not be delivered is recorded with its reason — no rule configured for that stage, or nobody staffed in that role. It is a report you can read, not a line lost in a log file.
The alert says which requisition, which quotation, which cylinder — and the link opens that record, filtered and ready. Not a list of four hundred rows with the answer somewhere inside it.
Not everything deserves a deadline. Fifteen of the seventy-one are informational: they appear in the feed, they never turn red and they never escalate. A system that shouts about everything gets ignored about everything.
Coverage
The interesting hand-offs are rarely inside a department. They are the seams between them — sales to production, the filler back to QC, inspection to the store — which is exactly where a spreadsheet has nothing to say.
Enquiry, quotation and its amendments, order, and sales review hand-offs into production and service.
Queued, completed, re-test created, sent out for refill, restart after service.
Stock reserved, movement tasks and transfers — mostly informational, one on a clock.
Requisition raised, pending approval, RFQ, vendor quotation, purchase order.
Goods receipt created, cylinder receiving accepted, refill inwarded.
Allocation pending, and routing on to production, QC or service.
Customer cylinders arriving, returns available, refills back from the filler.
Passed to final inspection, failed to rework, re-test created.
Sent to service, service completed, item service completed.
Item inward expected, cylinder inward pending.
Cylinder inwarded post-inspection, item quality inspection completed.
Ready for picking, packing completed.
Dispatch-ready and delivered notices.
Stale enquiry, expired quotation.
Manager review pending — the one that escalates to super admin.
Passed, and therefore dispatch-ready.
Cylinder queued for refill at a third-party filler.
Where it lands
The base channel, always on: a live counter in the header, a full inbox, and a per-zone overdue-and-escalated summary on the plant-flow command centre.
A distraction-free kiosk with large touch targets and an audible chime when a new move task lands — for the bay, where nobody is watching a browser tab.
Switched on per branch with your own sender identity, routed to the right department mailbox by role. Dispatch runs as a scheduled sweep and every send is stamped on the notification itself.
The channel is implemented and provider-agnostic. It is waiting on business API provisioning, and we would rather tell you that than list it as available.
On screen
Reminders
The hand-off rules cover work moving between desks. These cover the other kind of deadline — the one attached to a date rather than to a hand-off. Each runs on a schedule and each is switchable per branch.
Walks every open, overdue notification and bumps it to the escalation role. This is the job that makes an SLA mean something instead of being a decorative timestamp.
Flags enquiries that have gone stale past your configured day count, and quotations that have passed their validity date — before the customer rings to ask where their price went.
Groups overdue invoices by customer and sends the reminder, logging every attempt, so you can prove what was chased and when.
Bills accrued rent monthly against a billing watermark, so the same day is never billed twice and never quietly skipped.
Measures free stock — net of live reservations — against reorder point and safety stock, and raises the requisition itself rather than waiting for someone to notice the shortage.
Analyzers and machines approaching or past calibration surface on a due list. An instrument whose expiry was never recorded is reported as unknown — never as valid.
The fleet is banded into overdue, due within 30 days, due within 90, later, and not recorded. That last band is its own answer: a cylinder with no test date on file is not the same as one in date.
Releases stock reserved against demand that never materialised, so the shortage you are being warned about is a real one.
Vendor prices lapsing inside 30 days are listed with the day count, so a rate contract gets renegotiated before it silently reverts to list price.
If a document cannot be pushed to your accounts system after its retries, the accountant and the branch manager are told. A sync failure nobody hears about is the expensive kind.
The difference
The order sat for three days because the person who takes cylinder assignments was on leave.
With Fluids The task is addressed to the role, not the person, so whoever is on shift sees it — and at 48 hours it is on the branch manager's screen whether anyone was on leave or not.
Everyone gets a copy of everything, so nobody reads any of it.
With Fluids Eight roles, 71 rules, and fifteen of them deliberately carry no deadline at all. What turns red is a small enough set that red still means something.
We have SLAs. They are in the quality manual, and nobody has read it since the audit.
With Fluids The SLA is the row that raises the task. If it is not in the rule table it does not exist, and if it is, it is running right now.
A cylinder failed its gas test on Friday and production found out on Monday.
With Fluids A failed test tells production, QC, the branch manager and admin at the same moment, on an eight-hour clock, and raises the rework job itself.
The clock only starts when somebody logs in and notices.
With Fluids It starts at the event. A task raised at six on Friday evening is already eleven hours old when the plant opens.
An alert went to a role nobody holds at that branch, so it went nowhere.
With Fluids It is re-addressed to the branch manager, and the substitution is written down — so the report tells you which desks you have not staffed.
FAQ
Yes — that is the whole design. Notified role, additional roles, the SLA hours, the escalation window, the escalation role, the priority and whether the rule is on at all are all edited on an admin screen and take effect on the next event. There are 71 rules configured today and none of them is written in code.
Leave the SLA hours empty. The notification is still raised and still appears in the feed, but it never acquires a due date, never turns overdue and never escalates. Fifteen of the current rules are deliberately set that way.
They go to the branch manager instead of being raised against an empty role, and the substitution is recorded. The delivery report then shows you exactly which roles are unstaffed at which branch, which is usually more useful than the individual alerts.
In-app is the base channel and is always on. Email is switched on per branch with your own sender identity and routes to the right department mailbox by role, dispatched on a scheduled sweep, with each send stamped on the notification. WhatsApp is built and awaiting business API provisioning — we list it as not yet live.
Because a drop is a record. If a stage fires and no rule matches it, or a rule names a role nobody holds, that is written down with the reason and shown on a delivery report. Most systems lose that in a log; here it is a screen with a count on it.
Both, on one engine. The same mechanism that gives a failed gas test an eight-hour clock gives a quotation a ten-working-day follow-up clock that escalates to the branch manager after twenty. Separately, a scheduled job chases overdue invoices, and rental billing, reservation reclaim and e-way bill expiry run on their own clocks. Sweeps for expiring quotes, lapsing vendor prices, calibrations coming due and cylinders approaching their hydro-test date are built and run when you run them — we have not put them on a timer yet.
Next
The SLA engine is only useful because there is a real operation underneath it. Start with the cylinder lifecycle, or see how the whole thing is switched on and off per branch.