AVT for general practice: How do you choose the right one?

Ambient voice technology (AVT) is moving quickly from early adoption to a more familiar part of the conversation in general practice. It helps clinicians to focus on the patient rather than the screen, with the NHS England 10 Year Health Plan identifying AI scribes as a way to reduce the administrative bureaucracy and free up time for care.

But the best AVT on paper is not necessarily the right AVT for general practice.

Time saved matters, but so does what happens when the technology meets the reality of different consultations, clinicians and workflows. Does it remove work or simply move it somewhere else? How much checking does it require? What happens to consultation data? And what support is there once the technology is live?

Choosing the right AVT for general practice means looking beyond the demo and considering what clinicians and practice teams will actually have to live with day to day.

What should practices consider beyond time saved? 

Time saved, ease of deployment and integration with existing systems all matter. They are often the first questions asked when considering AVT.

But a headline saving of a few minutes per consultation does not tell you where that saving comes from, or where it goes. The more useful question is whether AI scribes remove work or relocate it.

If less time is spent writing notes but more is spent correcting them, the benefit may be smaller than it first appears. Equally, finishing documentation during the working day rather than after a clinic may have value beyond the number of minutes saved in an individual appointment.

Practices should therefore look at how AVT fits across the working day, not just how quickly it produces a note.

They should also consider the range of consultations clinicians actually have. A straightforward telephone review can look very different from a complex face-to-face consultation covering several problems. The right AVT for general practice needs to work across the consultations your team handles without requiring clinicians to change how they communicate with patients simply to accommodate the scribe technology.

Start, too, with the tool’s intended purpose. Is it designed to create a draft clinical note? Does it suggest codes for review? Does it generate correspondence? Understanding what the technology is designed to do, and where its role ends, gives practices a clearer basis for deciding whether it fits their needs.

How do you choose the right AVT for general practice? 

Choose the right AVT for general practice by looking at how it fits consultations and workflows, what it asks of clinicians, how patient data is handled and what happens after implementation.

A good starting point is to see the technology in the context in which clinicians will actually use it. 

• Does it work across face-to-face and telephone consultations?
• What does the clinician need to review before information reaches the patient record?
• How easy is it to correct or amend an output?
• Does it support the wider documentation created around a consultation?
• What support is available when clinicians have questions or something does not look right?

The review experience matters because the clinician remains responsible for reviewing AI-generated information before it is relied on for patient care.

That means successful implementation is not simply about teaching people how to turn AVT on. Clinicians need to understand its limitations, know what to look for and remain in control of the final record.

There is a human factor here too. As a tool becomes familiar and produces good results repeatedly, it can become easier to trust its output automatically. Practices therefore need to think about how meaningful clinical review is maintained once an AI scribe stops feeling new.

As Dr Dom Pimenta, CEO and founder of TORTUS AI, explores in how to buy AVT in 2026, assurance should not be treated as a one-off check at the point of purchase. AVT continues to change after implementation, so practices need to understand how performance is monitored, how updates are assessed and how concerns are handled throughout the life of the technology.

What happens after AVT goes live? 

Choosing AVT is not just a procurement decision. It is the start of an ongoing relationship with technology that will continue to change.

That matters particularly with AI.

With conventional software, an update might change what a screen looks like or add a feature. With AI, changes to models, prompts or functionality can also change how the system behaves.

The AI scribe a practice evaluates today may therefore not be exactly the same technology its clinicians are using six or twelve months from now.

Practices do not need to become AI experts themselves. But they should understand what happens when the technology changes, how updates will be communicated and the support available to them after go-live. If clinicians identify something that does not look right, who will investigate it and feed that learning back into the product?

The relationship with the supplier matters.

The right partner should be able to support implementation and ongoing use, explain changes clearly and give practices visibility of how the technology continues to perform, rather than leaving them to resolve concerns alone.

Related read: General practice workflow optimisation beyond go-live

Why does data sovereignty matter when choosing AVT for general practice? 

Where does a consultation go after the patient and clinician stop speaking?

That is an important question when choosing AVT for general practice.

Practices need to understand where sensitive consultation data is processed and stored, who can access it, how long it is retained and whether it is used to train AI models.

Traditional AI approaches can involve moving data into external platforms or shared AI environments for processing. In healthcare, this can raise questions about data access, governance and operational oversight.

These concerns are already being felt across the NHS. Research cited by HTN in why sovereign AI matters for the NHS found that 44% of NHS organisations see security risks and vulnerabilities as a key concern, while the same proportion lack confidence in protecting data from cyberattacks.

Sovereign AI takes a different approach, bringing AI capabilities closer to the data within environments where organisations can maintain greater visibility and control.

For practices, this can support clearer oversight of patient records, access permissions and audit trails. It can also make it easier to understand how AI is being operated, rather than treating data handling as a technical detail outside the practice’s control. Clear answers about how information is handled are also important for trust in AI in healthcare, for both patients and practice teams. 

TORTUS AI reflects these principles through UK-based data processing and storage. Audio is not retained, and patient data is not used to train AI models. This means Surgery Intellect, powered by TORTUS, can support clinical documentation without moving consultation information into a shared training environment.

Data sovereignty is not a replacement for wider information-governance checks. But as AI becomes more embedded in clinical workflows, knowing where consultation data goes, and what happens to it when it gets there, should be part of the buying decision.

Choose AVT for general practice that works beyond the demo 

A good AVT demonstration can show how quickly a note appears on screen. It cannot show a practice everything it needs to know about living with the technology.

The real test comes later: across different clinicians and consultation types, during a busy surgery, when something needs correcting, when the technology changes or when a practice needs support.

Surgery Intellect, powered by TORTUS, is an ambient scribe designed for clinical documentation in UK primary care. It supports face-to-face and telephone consultations while keeping the clinician in control of the final output.

TORTUS is the first AVT to achieve UKCA Class IIa certification, with external assessment of its quality-management and safety processes. That provides an additional level of assurance and supports safe AI in primary care, but practices should still understand how any AVT fits their own workflows, what it asks of clinicians and how it will be supported over time.

The right AVT for general practice is therefore not simply the one that produces a good note or promises the biggest time saving. It is one that works with the way your clinicians consult, reduces rather than relocates work, gives you clarity over sensitive patient data and comes with a supplier you can work with after go-live.

Once you know what your practice needs from AVT, the next step is testing what suppliers can demonstrate.