Good practice

Now, is it necessary? Yes.

In certain areas like life sciences, regulated industries, where there are GXP guidelines, generally better accepted principles, there it is almost mandatory to have some kind of a sign-off.

We have only integrated DocuSign because it's easy to work through it.

It could have been as easy as sending an email to someone and saying, please approve, and somebody says approved, and it comes back.

We needed, you know, people are like, I need to see where this has worked or, what are the use cases? Cause there was a lack of, I think for a little while in industry, some creativity because they were nervous about the fact that, you know, they would have to go through governance.

They would have to in a lot of cases, depending on what the use case was validated and meet the GXP requirements and sometimes codes of practice, right? Those are things that I think all around are things that people use as a really strong reason not to lean into AI.

And so what, I found was that firstly, we had to focus on the use cases.

I think that was the first thing that I have constantly really focused on to move us forward.

And so I think the projects that we, like I said, have had success in are because we were crystal clear on what we wanted to do.

And I think, you know, I've talked about content generation, the process optimization and the system integration that was totally facilitated in our case by using our own, you know, sort of ecosystem to tie these together based obviously on a large language model, then with a large framework put around it.

And then to actually validate it and say, these are the rules that we work by within the GXP space, within our local regulations, our laws, and making that environment then really fit for purpose so that we could access information, then eventually create information, then eventually make really good data and information.

But even that too, you know, again, going, we expanded that as well now, and we're doing more projects with this as call centers.

And some of these get very specific.

I won't go into things like NASBA requirements, but it gets very complicated at GXP, how many questions you can ask in an hour, et cetera.

So we're driven by that because we have to.

These are business requirements.

Wenn wir die jetzt noch auf quasi transparent und anwendbar machen, oftmals bei 'nem matischen-

Aber das Spannende ist ja, find ich auch, dass Pharmaunternehmen mit GXP waren ja SOPs, Standardanweisungen.

Da stecken ja immer ganz viele Regeln.

Deswegen genau, aber das waren-

I bet.

And the expert tier also demands mastery of GXP AI compliance.

Okay, GXP being the good practice guidelines?

Correct.

And then I made a list of every requirement that I need this data to fulfill and the format I need.

So it was a GXP data.

Like I said, like nothing should be changed.

Like if there is anything you don't understand, like put a question mark, like something like this.

Yeah.

And I think, honestly, this is probably where the fact that it is clinical and it is covered by all the regulations, both on IT systems like GXP, computerized system validation, data integrity and all that stuff.

It also makes it, the rigor, there's so much rigor that goes into ensuring consent and structure and truth, which also means that it's sort of the value of it just increases a lot because we know this data, we know so much about it, we know what we're allowed to do with it.

Well, to some degree, depending on the updated consent form.

Then you go into your, I don't know what now NovoNautics is using, but we use something called LSAP.

It's a statistical computing environment where we do the programming.

So you find something, you bring it to LSAP, and then you can do it in a more GXP compliant way,

follow it up.

They give us 24 seven, you know, monitoring.

So it's SOC two complete, make sure that everything in our, our operation is SOC two compliant HIPAA 21 CFR part 11 GXP, et cetera.

And I think one thing that's cool about their company, they actually have a customer facing portal.

So if I sold to your site, let's say, you know, you can log on to Vance and say, OK, let's make sure the intakes, you know, got in their eyes and crossing their T's.

And our regulators are getting very smart, actually, and asking these really good questions and challenging us.

And it really is part of that new exercise of showing that we can demonstrate our inspection readiness, our adherence to the GXP standards and many more, even our codes of practice and everything else.

We really are being asked to understand our solutions and that when we are using AI, that we are making sure that it's still whatever we create, it's always going to be accurate information, balanced information.

Obviously, for the medical teams, it's non-promotional in nature.

I say that's still part of traceability.

And you're able to articulate to a regulator very clearly.

You and even your team members are able to say, what actually are you doing and what is the AI doing? And do you understand that? And I think on the other side of it is like any GXP system on the back end, can we tell who did what when we check it? Can we actually go back and can we reproduce an audit trail of sorts? And I think the other part of it is really, again, Do we understand the logic that the AI is using? I think this is really important.

And what's the source of truth? And so those all, I think, in combination are really critical and are really being asked now by our regulators, which is good.