That's why Medivox is building its own infrastructure for the entire value chain

In recent years, developments in artificial intelligence have progressed at a pace few could have predicted. Where AI models used to be heavy, costly and completely dependent on global cloud providers our analysis now a clear shift:
The models will be faster, easier and far more efficient - and can increasingly be run locally, on dedicated infrastructure.

This is not just a technological improvement.
It's a strategic choice for the entire AI industry - especially in healthcare.

Our analysis: why the industry is still cloud dependent

Our analysis shows that it is not common that AI solution providers in the healthcare sector process data themselves.
On the contrary, the most widespread setup is still based on:

- international cloud solutions

- Ready-made AI models delivered as a service

- third-party processing of both audio and text

This is directly reflected in prices in the market. When every transcript and journal draft is processed by external providers, with ongoing API costs and cross-border data traffic, both cost levels and legal complexity are high.

The reason is often simple:
Until recently, it has been technically demanding and costly to operate AI models locally.

Models change the rules of the game

Based on our analysis of today's AI models, we now see that this picture is about to change. The models have become:

- significantly faster

- far more resource-efficient

- better suited for local operation on modern servers

This opens up for a completely different architecture - where AI can be moved closer to the data, instead of data being sent out into the world.

It is this development that Medivox has chosen to take the consequences of.

Medivox: a deliberate break with industry practice

Ever since its inception, Medivox has had a clear goal:
AI in health should be used with the best possible control over the data being processed.

An important step on this journey has now become a reality:
Today, Medivox processes all transcription itself - on its own infrastructure - for all our customers.

This is a major technological and strategic step forward.

In practice, this means:

- audio data does not leave Medivox's infrastructure

- we are not dependent on international cloud solutions for transcription

- we have full control over where and how data is processed

This is still uncommon in the industry - but in our opinion it is absolutely necessary.

Lower prices - better GDPR profile

The fact that we now process transcription ourselves has had two clear consequences:

1. We have been able to cut our prices drastically
Removing external processing costs also reduces the ongoing cost per consultation. This has enabled us to offer significantly lower prices than many comparable solutions.

2. Customers get a much better GDPR profile
Less data sharing, fewer subcontractors and shorter processing chains mean less risk - and easier documentation of compliance.

Where we currently use subcontractors for parts of the text processing, this will happen only after pseudonymization. Common personally identifiable information is removed before the text is processed further.

Next step: all text processing on your own servers

Based on our analysis, the next step is already clear:
also full text processing is gradually moved to Medivox's own infrastructure.

Our long-term direction is clear:

- no subcontractors for AI processing

- all processing with AI on your own servers

This is the way the industry needs to go

Our analysis shows that more and more players are now moving in the same direction:
from general cloud solutions to local, controllable and specialized AI setups.

In the healthcare sector, this is a competitive advantage, both in terms of costs and compliance.

Medivox's promise

Medivox doesn't just build functionality.
We build long-term trust.

By owning the infrastructure, controlling the data flow and taking responsibility for the entire processing chain, we make it possible to use artificial intelligence in health - in a way that is more affordable, more secure and more predictable.

This is how responsible health AI should work.

Mats. F. Evensen
General Manager
Medivox