Skip to content
VOMEDIQECG IntelligenceAI-Assisted ECG Analysis
Request access
The productHow it worksWho it is forPricingSafety & limitsSign in

Who builds this

A system that says what it does not know.

Clinical software is unusual in that its most valuable behaviour is often its least impressive one: refusing to answer. Everything below is a decision that made this build slower to produce and calmer to be accountable for — and each one names what it costs.

There is no customer list on this page, and no funding announcement, because neither would help you decide anything. What would help is knowing how the people behind the software make decisions, since that is what you will be relying on long after this conversation — when something in it has to change and nobody from here is in the room. So this page is written as an engineering position: each commitment, and the thing it gives up.

The engine decides. A model may only describe.

Every measurement, finding and severity comes from a versioned deterministic rule engine reading the stored samples. A language model, where a deployment enables one, turns validated findings into readable prose. It is never asked what the ECG shows.

What it costs: This is slower to build and worse at sounding impressive. A generative model asked directly for an impression would produce confident prose about any trace in milliseconds — including the traces it is not entitled to say anything about.

An unusable recording produces no interpretation.

Automated interpretation is published only when the recording passes quality validation. Where it does not, the result is the explicit statement that interpretation was withheld — which is a different fact from a normal ECG and is presented as one everywhere it appears.

What it costs: A system that always produced an answer would look more capable on a dashboard, and a department measuring "interpretation coverage" would see a better number. Withholding is the less impressive behaviour and the correct one.

A person decides, and the record shows which person.

Review, override and sign-off are separately permissioned acts. A signature is bound by digest to the exact interpretation it covers and to the professional who gave it, and a disagreement is stored as a disagreement rather than folded into an edit.

What it costs: There is no configuration in which the software can be left to act on its own, and no "auto-approve" switch that a busy department could set. Some of the friction here is deliberately irreducible.

The limits are published next to the claims.

The public site states what the build does not do — no alert about a patient reaches a person by any channel, nothing renews automatically, no invoice or refund exists, and no accuracy figure has been established — and the software reports the same absences from its own status endpoint. The one outbound message the build does send is operational rather than clinical: notice that a machine is failing or has gone silent. That is possible only where the deployment has a transport, and where it has none the product says nobody was told rather than counting it as sent.

What it costs: A conventional marketing page would leave those to be discovered during an evaluation, and would convert better. This one does not, because a hospital that discovers an absence after signing is a hospital that stops believing the rest.

Clinical data stays inside the deployment.

The content security policy permits no third-party origin other than a payment provider on the billing screen. There is no font host, no icon CDN, no analytics script and no error-reporting service, so there is no third-party request to disclose.

What it costs: Convenience tooling that most software depends on is unavailable here, and several things that would take an afternoon take a week. The trade is that the network diagram has no dotted lines out of it.

What a buyer should expect

An early product, described as one.

This build is at the stage where the software is more complete than the evidence around it. That is worth stating plainly, because it changes what a sensible next step looks like.

The engineering is ahead of the paperwork

Validation against a clinical dataset, independent security testing, a quality management system and regulatory registration are all absent. None of them is a formality and none of them can be replaced by a better product page.

Evaluate it on its behaviour, not its claims

Every capability described on this site corresponds to something in the build that can be exercised: the quality gate withholds on an unusable recording, a signature stops verifying when a finding changes, the audit chain breaks if a row is edited. Those are checkable in an afternoon.

Start on synthetic data

The hosted deployment serves recordings generated by the system’s own synthetic ECG generator. You can see the whole reading experience without your own patient data being involved, and without involving your information-governance process to get there.
IT, procurement and deployment The product, screen by screen

The four commitments

Stated once, and stated as claims.

The same four that appear on the home page, in the same words. A commitment that appears in two different phrasings is two different commitments, and the reader is entitled to check one against the other.

The engine decides. The model describes.

No finding in this system originates from a language model.

Measurements, findings and severity are produced by a versioned deterministic rule engine reading the stored samples. A language model is used only to turn validated findings into readable prose — it is never asked what the ECG shows, and a statement it produces that is not traceable to a validated finding is a defect.

Quality is a gate, not a footnote.

An uninterpretable recording produces no interpretation, not a cautious one.

Automated interpretation is published only when the recording passes quality validation. Where it does not, the result is the explicit statement that interpretation was withheld — which is a different thing from a normal ECG, and is presented as such.

A person is in the loop, by permission rather than by habit.

Review, override and sign-off are separately permissioned acts, and disagreement is recorded.

A signature is bound by digest to the exact interpretation it covers and to the person who gave it. An override is stored as an override for ever: the automated finding, the clinician’s finding and the reason, all attributable.

Nothing is asserted without the evidence for it.

Every claim on this site about a capability maps to something the software does.

The limits stated on these pages are not modesty. Automatic renewal, email delivery, invoices and refunds are genuinely absent, and the product reports them as absent in its own health probe rather than leaving them to be discovered.

Who to contact

A company, an address and a number.

A page about who builds this that did not say how to reach them would be an essay. This is the part a department actually needs: one supplier, reachable three ways, and a clear statement of what not to send them.

AI Algo (S) Pte Ltd.

Telephone
+65 9636 9376
  • Reach the supplier for anything about the software: how to bring recordings in, why a file was refused, how to attach a machine, whether a setting is right, or a screen that does not behave.
  • Do not route a clinical question through the supplier. A result that looks wrong, a finding you disagree with, or anything concerning a patient belongs to your own clinical governance and to the clinician who signed it.
  • An emergency is an emergency. Nothing in this product, and nobody who supports it, is part of a clinical escalation path.
  • The System & Integration screen is generated from this host’s running configuration. Sending that report with a question answers most of it before it is asked.