What faster development means for QA in healthcare

C the Signs¹ is an AI-powered platform that helps clinicians spot cancer risk earlier and is used in 1,500+ GP practices across the UK².

This is what Lorenzo had to say: 

AI is changing how quickly we build software, and we're seeing this already at C the Signs. With tools like Claude and Cursor, I'd estimate our development throughput has increased by around 1.5 to 2 times.

Of course, the expectation is that if development is moving faster, we should be able to deliver more. But the whole process doesn't automatically become faster just because you can write code more quickly.

You still need to test it. And in healthcare, there are some things you really can't afford to get wrong.

When QA becomes a bottleneck

We automate a lot of testing already, and there are more opportunities to use AI to make that work faster. But I don't think AI-driven QA is there yet for critical healthcare flows; those still need people.

In healthcare, you can have hundreds of different flows through a product. Most of them might work perfectly, but you only need one important flow to fail for it to become a patient safety issue.

So, if development suddenly becomes much faster, QA can become the bottleneck. You're producing more software, but it still needs to get through testing before you can release it.

For product teams, I think that's something we need to start thinking about now. If development is moving twice as fast but you can't release twice as fast, you haven't solved the whole problem.

Where human judgement adds value

This doesn't mean manually testing everything.

Automation is essential when you're dealing with a lot of different flows and releasing regularly. AI can help with more of this too. The question is where we're comfortable relying on it, and where we still need a person involved. For me, critical healthcare flows are still in that second group.

A tester can explore different scenarios, question unexpected behaviour and recognise when something that looks like a small issue could have a much bigger impact. That's where the value of human judgement comes in.

The balance will change as AI improves. There will be things we use people for today that we won't need them to do in the future. That's a good thing if it means experienced testers can spend more of their time on the areas where they add the most value.

I don't see it as people versus AI or automation. They enable each other. Use the tools to create coverage and efficiency, and use people where context, judgement and risk matter.

The wider cost of finding problems late

There’s another reason why getting QA right matters in health tech.

If you find a bug late, you don't necessarily just fix it, test it again and release. Depending on the product, you might have to repeat your regression testing. Then there can be clinical testing and clinical safety review. You may need to update technical documentation or your technical file. If that has to be reviewed or submitted again, you've added even more time.

That's the part people sometimes miss when they talk about the cost of rework in healthcare. The cost isn't only the engineering time needed to fix something. It's everything that has happened around that software and might now need to happen again.

And usually, you're working towards a date. You have a release planned, an audit coming up or a regulatory requirement you need to meet. Those dates don't move just because you've found a problem.

Finding important issues earlier gives you more options and makes it easier to deal with them before they affect everything else.

Evolving QA alongside development

I don't think we know yet exactly how much AI will change software development or QA. But we're already seeing enough to know that product teams need to think about the whole delivery process, not only development.

If AI allows us to build more software, QA needs to be able to deal with that increased throughput without losing the scrutiny that healthcare products need.

For me, that means making the best use of everything available to us. Automation gives us coverage and helps take care of repeatable work. AI will make more of the process efficient. And experienced testers can focus their attention on the areas where context, judgement and risk matter most.

The balance between those things will keep changing as the technology improves. But the principle is fairly simple: writing software faster is only useful if we can still release it safely.

¹C the Signs is not a Zoonou client, and Lorenzo’s views are not an endorsement of Zoonou or its services.

²https://www.cthesigns.com/media-centre/defying-time-how-c-the-signs-is-transforming-early-cancer-detection-across-the-nhs