Near-infrared imaging
A dedicated NIR camera reveals the subcutaneous vein network that's hard to see or feel by eye.
Technology
Three ways of seeing, one clear target.
VeinMate Assist fuses complementary sensing with on-device intelligence and an independent safety layer — engineered with production-grade components from the start.
How it works
From live image to recorded session.
The device does the seeing; the clinician does the deciding and the doing — the same four steps every time.
Every step of AI output is advisory; the clinician stays in command from first frame to final access.
Sensing
Complementary ways of seeing the vein.
Ultrasound depth sensing
Adds vessel depth and width, so guidance accounts for the third dimension, not just the surface picture.
Single-use guide cartridge
The patented, sterile consumable that references guidance to the patient and keeps each access clean.
How it flows
How the layers fit together.
Sensing, intelligence, and safety are separate layers that hand off in sequence. Every AI output along the way is advisory — the clinician stays in command from first frame to final access.
1
Capture
The device captures near-infrared and ultrasound signals together, combining the surface vein picture with vessel depth and width.
2
Segment
On-device models segment the vein network from the fused signals — no live cloud connection is involved during the procedure.
3
Highlight
The system highlights an optimal access point and shows a confidence indication, so the clinician can see how strongly the guidance is supported.
4
Confirm & perform
The clinician reviews the guidance, confirms or overrides it, and performs the access. The device images and guides; it never inserts on its own.
5
Record
The session is recorded as a structured, anonymised record for quality improvement and later review.
Every step of AI output is advisory. It informs the clinician's decision and never replaces it.
AI guidance
On-device intelligence.
VeinMate Assist is edge-first by design. Vein segmentation and insertion-point detection run on the device itself, so guidance does not depend on a live cloud connection while a clinician is with a patient. The models are optimised for edge inference and developed with attention to how they perform across skin tones and patient types.
Edge inference
Vein segmentation and insertion-point detection run on the device itself, on a production-grade compute module — no dependence on a live cloud connection during a procedure.
Built for everyone
Models are optimised for edge inference and developed with attention to performance across skin tones and patient types.
Advisory, always
All AI output is advisory: it informs the clinician's decision and never replaces it.
Performance figures quoted in briefings and technical materials are engineering targets under active validation, not published clinical results.
Skin-tone equity
Designed for skin-tone equity.
Vascular access shouldn't be easier for some patients than others simply because their veins show up more readily under infrared. Guidance is designed to work across the full range of skin tones — not just those that image easily — so more patients benefit from the same standard of assistance. This is a qualitative design commitment; performance across patient groups is part of ongoing validation.
Safety by design
Safety architecture.
A dedicated safety microcontroller runs independently of the main compute module on an isolated power rail. Its only job is safe-state behaviour — the guarantees that hold even if the higher-level system misbehaves. Safety is a separate layer, not a feature of the software stack.
- Independence — the safety controller is separate from the main compute module, so it isn't compromised if the higher-level system fails.
- Isolation — it sits on its own isolated power rail rather than sharing power with the rest of the system.
- Safe-state defaults — its single job is to hold the device in a safe state, the behaviour that must always apply.
- Clinician in command — the clinician performs and remains in control of every access at all times.
Records you can trust
Data & audit.
Every session produces a structured, anonymised record: outcome, timing, and device telemetry. That trail supports quality improvement today and clinical-record integration tomorrow — with data security built in from the ground up.
- Structured records — each session captures outcome, timing, and device telemetry in a consistent, anonymised form.
- Quality improvement now — aggregated records help teams understand and improve how vascular access is delivered.
- Integration later — the same structured trail is designed to support clinical-record integration in future.
- Security built in — data protection is designed into the record from the ground up, not bolted on afterwards.
On the roadmap
Honest about what ships and what's ahead.
Photoacoustic tomography (PAT) is a roadmap technology for deeper and richer vessel characterisation. It is not part of the current device. We're honest about the line between what ships and what's ahead.
PAT is roadmap-only. It is not a current capability of VeinMate Assist.
Common questions
Technology, answered plainly.
Does it need the cloud during a procedure?
No. Inference runs on-device. Vein segmentation and insertion-point detection happen on the device's own production-grade compute module, so guidance does not depend on a live cloud connection while a clinician is with a patient.
Are the accuracy numbers proven?
No. Any performance figures we share are engineering targets under active validation, not published clinical results. VeinMate Assist is an investigational device and is not approved for sale or clinical use.
How does it connect?
Over secure Wi-Fi. Connectivity is used for tasks such as transferring anonymised session records — not for running guidance, which stays on the device during a procedure.
Is PAT in the product?
No. Photoacoustic tomography is a roadmap technology only. It is not part of the current device or a capability of VeinMate Assist today.
What components are used?
Production-grade components throughout — from the compute module to the sensing and safety hardware. The device is engineered on production-grade parts from the start.
Partner with us
Go deeper in a technical briefing.
See how the sensing, AI, and safety layers come together. We'll walk your team through the architecture and where each engineering target stands in validation.
Request a briefing