Benefits
Better for some of these people than others.
Eight roles, one at a time: what changes, and what it costs them. No percentages anywhere, because no laboratory runs LabFlow yet.
Benefits pages usually lie by aggregation.
They average a gain across a whole laboratory, attribute it to the software, and print it as a number. The technician who now types more and the manager who now sees more both disappear into one figure that neither of them recognises.
So each section below names one person, says what changes for them, and says what it costs them. Where the cost is real it is written down, because a benefits page that has no costs on it has not been read carefully enough by the person who wrote it.
Technician and senior technician
You stop being the place where the reference interval is remembered.
The interval is resolved for the patient in front of you, by age, sex and method, and the comparison is recorded rather than performed in your head at the end of a shift. A value that moved implausibly since the last one is flagged before you open the record, not after you have signed it.
The worklist is a real queue built from ordered tests that have no result yet, so what is waiting and how long it has waited are properties of the data rather than of a whiteboard.
When a release is refused, the refusal names the rule. You never have to work out why the button did nothing.
- Ranges resolved, with the interval named
- Delta checks run before you look
- A worklist that ages visibly
- Refusals that quote the rule that refused
- Offline entry that reconciles rather than overwrites
- A rejection now needs a reason, chosen from a list, every time
- An override is recorded with your name on it, which is the point and is still uncomfortable
- Nothing here reduces the number of tubes that arrive
Pathologist
The question you are asked six months later is answerable.
Almost every difficult conversation about a laboratory result is a question about the past: what did the report say when the clinician read it. A system that overwrites cannot answer it.
The record now shows the correct value
Which is the one thing nobody is asking about. The value that was acted on is gone, the report that carried it has been silently reissued, and the reconstruction is done from memory and email.
A status called amended tells you a correction happened. It does not tell you what was believed at the time.
Both rows exist, and both are true
Version 1 is still released, because it was. Version 2 points at it, carries the reason, and names you. The issued report is a snapshot that was not rewritten, and the recipient of the original was told.
The reconstruction is a query, not a recollection.
What it costs you. Amending is heavier than editing. You supply a reason, the recipient is notified, and the old row stays visible to anyone reading the record. Every one of those is deliberate friction, and on a busy afternoon it will feel like friction.
Laboratory manager
Four things you currently chase by asking people.
Each of these is a question a manager asks out loud several times a week. Each becomes a query against data that was recorded as a side effect of the work, rather than as an extra thing somebody had to remember to do.
Where is that specimen
Custody is a chain of rows: collected, in transit, received, stored, and every hand between. The answer includes the exceptions, so a temperature excursion recorded on reconnect is part of the answer rather than a separate conversation.
Cost: somebody has to scan or confirm at each hand-off. A chain with gaps is a chain that answers the wrong question confidently.
Why was that result late
Turnaround is measured from receipt, which is the moment the laboratory took responsibility, and the waiting is visible per test rather than per batch. A rejected specimen keeps its row, so a late result caused by a redraw explains itself.
Cost: the number will sometimes be worse than the one you quote today, because today's number is measured from a more flattering moment.
Was quality control in control when that ran
The control run is bound to the patient results produced beside it. A question about one result reaches the control that was in force at the time, with the rule that passed or failed named.
Cost: an out-of-control run now blocks release, which is the correct behaviour and will occasionally stop work at an inconvenient hour.
Who did that, and when
The audit log is written by the same path that performs the action, so it cannot be partially on. Overrides, amendments, releases and permission changes all leave a row with a person against it.
Cost: your staff know they are logged. That is the intended effect and it changes the atmosphere of a laboratory, for better and occasionally for worse.
Phlebotomist
A basement with no signal stops being a problem you carry back.
Seventeen custody actions work with the network off. You collect, label, record the condition and move on, and the queue reconciles when the device reconnects. Conflicts are surfaced for a person to resolve rather than silently resolved by whichever write arrived last.
The accession number is generated by the database when the specimen is received, so nothing depends on your device inventing an identifier that a second device might also invent.
custody actions that queue on the device and reconcile on reconnect
What it costs you. The application has to be installed and signed into before you leave, and an unsynced device is a real risk that somebody has to watch. Offline is not the same as free.
Ordering clinician
Four things that arrive with the number.
A result is only useful with its context attached. Sending the value alone moves the interpretation onto whoever opens it, usually at the least convenient moment.
The interval that was used
Not a generic normal range. The one selected for this patient's age, sex and the method that produced the value.
What changed since last time
The delta against the previous result, and whether it crossed a threshold that a person had to look at.
Whether it has been amended
If a result you received is later corrected, you are told. That is a communication obligation, not a preference.
Whether it is final
Preliminary results are labelled every time they are shown. A preliminary that looks like a final is a patient-safety defect.
What it costs you: a released report downloads as an HL7 v2 or FHIR file, and getting that file into the system you already use is yours to do. There is no live feed.
One timeline, even when two laboratories were involved.
A patient is a person and not a visit, so the record follows the person across every laboratory that runs LabFlow rather than fragmenting into one account per building.
What it costs you: a laboratory has to be running LabFlow for any of this to apply to you, and today none is. This section describes the design, and it is written in the present tense because the code is, not because the situation is.
Laboratory owner
Being listed is a review, and that is the value of being listed.
A directory anybody can add themselves to is worth nothing to the patient reading it, and therefore worth nothing to the laboratory in it. Five states, with a person in the middle.
What it costs you. Review takes as long as it takes, because the LabFlow team does it by hand. A rejected application is told why and can be resubmitted, but there is no service level attached to any of this and none is claimed.
Platform administrator
The administrative surface is a product, not a back door.
It is designed, responsive, themed and audited to the same standard as everything a laboratory user touches. An admin panel treated as an afterthought is one that becomes unusable on the phone its operator actually has with them.
Review queues that hold state
Laboratory applications arrive in a queue with their documents attached and their history intact. A rejection carries a reason the applicant can read.
Roles that are enforced, not decorated
Granting a role changes what the server allows, not only what the interface shows. A control that hides a button while the endpoint stays open is the failure this is designed against.
An audit log with no exceptions
Including the administrator's own actions. A log that omits the most privileged user is the log you cannot use in the investigation you needed it for.
No percentages. Here is what to measure instead.
Every improvement figure on a laboratory software website is measured by the vendor, in a laboratory you cannot inspect, against a baseline you were not shown. These four are measurable in your own laboratory with the system you have today, which makes any later comparison yours rather than ours.
Receipt to release, per test
Measured from the moment the laboratory took responsibility, not from when the analyser finished. Record the median and the worst decile, because the worst decile is what generates telephone calls.
01
Time to answer a custody question
Pick five specimens from last month and time how long it takes to establish every hand they passed through. That number is usually the most surprising one in the exercise.
02
Amendments where the recipient was told
Not the amendment count. The proportion where you can demonstrate that whoever received the original was informed, and demonstrate it from records rather than from memory.
03
Releases against an out-of-control run
Today this may be unmeasurable, which is itself the finding. If you cannot count them, you cannot claim there are none.
04
If you measure these and LabFlow does not move them, that is worth telling us and it is worth publishing. The inbox is the same one.
The ledger
What does not get better, and in two cases gets worse.
A benefits page with nothing on this list has not been written honestly. These are the ones worth knowing before a decision rather than after one.
The same number of specimens arrive and the same number of people are needed to work them. LabFlow changes what those people can prove afterwards, not how many of them you need.
Nothing in software makes an instrument faster. Where turnaround improves it is because waiting is visible and hand-offs are recorded, not because the chemistry moved.
A rejection needs a reason. A custody hand-off needs a confirmation. An override needs a justification. Every one of those is a keystroke that did not exist before, and the value of the record is exactly the sum of those keystrokes.
An enforced release gate refuses. That is what an enforced gate is. If quality control is out of control at 02:00, somebody will be repeating a run at 02:00 rather than releasing and dealing with it in the morning.
An intact audit trail and versioned results make evidence easier to produce. They do not pass an assessment, and no vendor can promise that they will. The Part 11 validation documentation package is planned work and is not shipped.
Invoices, payments and claim records are built, and a claim’s denial and appeal are recorded by hand. No claim generation, no remittance parsing, no statements. Whatever you use to submit claims today, you will still be using it.
One team, one time zone, no rota. A laboratory that runs overnight should weigh that seriously, and it is on this page rather than discovered during an incident.
The useful next step is a conversation about one bench.
Not the whole laboratory. One discipline, one worklist, one set of reference intervals, and the awkward workflow that every laboratory has and no vendor demo ever covers.
What is actually built is on the features page. What it might cost is on the pricing page, including why no figure is printed there yet.
There is no trial to start and no demo tenant to sign into, so this site does not offer either. Both would be a button that does not do what its label says.