Executive Interview with Oliver May, SynSilico

0:06
Hi, everybody.

0:06
And welcome.

0:07
My name is Lucia.

0:08
I'm a Senior Digital Content Editor at Oxford Global.

0:12
Today, I will be interviewing Oliver May, a Managing Director at SynSilico, a startup company using AI to accelerate drug discovery.

0:20
So welcome, Oliver, and thank you for joining us today.

0:24
So I guess to kick things off, could you tell the audience a little bit about the problem that SynSilico is trying to address in the drug discovery space and what drew you to the company in the 1st place?

0:38
OK.

0:39
So what we and I will see is that in the discovery field, in the farmer field, there's a lot of question about how can we improve the efficiency of identifying your leads and pushing them through the pipeline to lend a drug.

0:56
So it's all about the efficiency of developments.

1:00
And what we wanted to do is use AI as one tool to rapidly solve through options and then apply computational tools next to it to at the end of the day, come out with less experiments, better experiments, more precise experiments to help the efficiency to improve in in the total process.

1:26
Thank you.

1:26
And I remember from so in our last conversation, you mentioned that your company is a joint venture combining computational expertise with life sciences and material science.

1:38
So I guess how does this combination uniquely position you in the AI drug discovery space?

1:46
Yeah.

1:46
So I think it's important to understand that the main knowledge is need to be combined to make progress in that space.

1:55
So if you only know how to use a computer and you don't know how to touch the result from your, let's say, protein engineering background or from your drug design background, you will lend perhaps with predictions that are meaningless.

2:11
And So what we try to do is we have a very strong AI and computational background from one company and another background for protein and life science domain knowledge.

2:26
And we bring this together in one company under one roof.

2:30
And what we also do is we bring also wet lab capabilities together because that is where things are proven.

2:39
And yeah, so this is we combine all of that in our one company from the background of our model companies with which come from completely different domain areas.

2:51
Yeah, I think having the multidisciplinary approach is very effective in a space like this.

2:56
And I guess we also touched on the fact that Sin Silico has three different pharma platforms.

3:02
Could you maybe discuss these in a bit more detail with the audience?

3:08
Yeah.

3:09
So I think the first we tried to put together was a small molecule discovery pipeline where we made use of the latest protein AI models to understand the structures and some aspects of dynamics of of structures induced filter aspects.

3:27
And the other part was that we had AI models for predicting binding energies.

3:34
So with that you we could automatically create a pipeline, you know, that can solve through huge numbers of small molecules and see how can they bind and what are the best molecules that bind best and then create inhibitors for that pipeline.

3:53
What we have seen there is that let's say the synthesis of those small molecules is giving us a challenge in terms of timelines plus the other part, the molecule dynamics part was not well enough developed in in that pipeline.

4:14
And that's why we've started to add molecule dynamics elements to that pipeline.

4:20
And we are focusing now in our second pipeline on large molecules, might sound a little bit strange because they are more complex.

4:29
But for us with our background in protein engineering and protein synthesis, we can immediately synthesise the designed molecules and make them available.

4:39
So this is speeding up that whole trajectory from elite and then designing an inhibitor against that lead and we can rapidly go down to the lab and provide those elite candidates.

4:55
1/3 pipeline, which is more a generic optimization tool is making use of Bayesian algorithms that promise that they can take insights from previous round of experiments into account and suggest better experiments in the next round.

5:14
And overall what it does, it's reducing the number of experiments.

5:18
If you have a complex optimization task in hand, it is not only for discovery, but also in in mainly for development activities.

5:29
So if you develop, for example, a recipe with that needs to stabilise the compound, then you can use such tools to rapidly define better recipes.

5:42
And this is more generic optimisation tool that I sent it for all kind of different applications that's.

5:50
And which of those projects did you enjoy working on the most, would you say?

5:56
I mean, I'm, I'm a protein engineer, so I have a little bit, yeah, I'm very curious how we can move the needle in the protein design space by AI.

6:09
And yeah, but it's not about my fun and my curiosity.

6:13
So at the end of the day, customers need to see the value of it.

6:18
And then I get joy in recognising that there's value in.

6:22
I have to say that the all three platforms received quite some attention.

6:28
Yeah, it's good to hear.

6:30
And how do you balance speed with efficiency and precision, I guess when using your platforms, especially with all the across all the different stages of drug, drug discovery?

6:46
Yeah.

6:46
So AI is very good in and very fast in solving through a large number of molecules.

6:56
But when it comes to defining and ranking the very best molecules, very often there's a compromise.

7:04
So this speed that AI brings, you know, there's a compromising precision.

7:09
And this is exactly where we also have to use a little bit more computational demanding tools to get this precision, but we can apply them on lower numbers of potential candidates that come out of an initial AI campaign.

7:25
So more or less, we combine speed that AI brings with precision that molecular dynamics and physical calculations provide us with.

7:37
Thank you.

7:38
And what types of collaborators or pharma partners do you expect to benefit from your unique approach?

7:47
Yeah.

7:47
So I think we are happy with both small companies, but also large companies.

7:53
Small companies often lack let's say the capacity and sometimes also the capabilities in the AI space and the design space.

8:04
So this is certainly very valuable for smaller companies to have a partner at hand that can execute on those AI supported design and screening programs, but also for large companies that always have large number of people in the background with very specific knowledge.

8:27
I think those companies also appreciate to have flexibility.

8:31
So if they are internal team is already fully busy with all the highly prioritised project but they want to test other things next to it, then service companies like ours are coming in quite handy for those large companies as well.

8:49
So it's both small and large companies that we serve.

8:56
Yeah.

8:56
And I think also from our last conversation, we kind of discussed a bit about how commercially raising awareness and building relationships can be tough initially for start-ups.

9:09
So I guess what strategies have helped you guys so far build trust with your potential pharma partners?

9:17
Yeah.

9:18
So if you're not directly from that domain that you are serving and you need to build networks, relationships, and they are things like conferences that you offer, yeah, are super helpful because you get in touch with the experts that can judge best what you're doing.

9:36
And those conversations are simply, yeah, so valuable that this is also the approach that we take.

9:43
And of course, just talking to someone there does not provide you with anything if you have not data to discuss with those experts.

9:56
And so it's the data that you have and the conversation around it that you need to go through and then you discover what experts see as a valuable offering that you have.

10:09
So, so we use conferences to build our networks to probe our value proposition, use that feedback and still tailor our offerings.

10:18
This is this is our approach that we are taking.

10:22
Very nice.

10:23
I'm glad bad to hear that you find the conference is useful.

10:26
And I guess are you prioritising specific projects in specific therapeutic areas at the moment or are you taking a more broad approach towards this?

10:41
So let's say those two platforms, the small molecule and large molecule design platforms, they don't target a very specific therapeutic area.

10:54
It all boils down to identifying inhibitors or interfering with with small and large molecules.

11:05
And this this is valid for all therapeutic areas, even for very distant things like gene therapies.

11:14
You also need to understand the network of interactions that you are manoeuvring.

11:22
Even if you just no got a chain, it comes with significant changes in networks.

11:26
And yeah, using antibodies or our design approach can help to understand the underlying mechanism.

11:35
So even such distant fields could also benefit from what you're doing, of course.

11:42
And then for investors looking into AI drug discovery companies, do you think they have any maybe common misconceptions about building businesses in this area that you would like to maybe address in this conversation?

12:01
Yeah, I mean our investor, it's known that it's milieu partners.

12:07
They have a strong long history in pharma developments.

12:11
So they know the field very well and they are not.

12:16
So they won't believe if I over promise to deliver now a truck within few years.

12:26
Yeah.

12:27
And so, but this is exactly where some investors they hear a story but cannot judge the background too much perhaps.

12:37
And then companies can easily over promise what you can do in pharma with AI and how fast everything works and how perfect everything is.

12:46
So I think this is not happening for us because we have knowledgeable investors and yeah, we also don't want to over promise what AI can do because we know it is limited and we tell our investors that it's limited.

13:03
But not everyone is so clear about what things can be done and cannot be done.

13:08
Yeah, very good message.

13:11
And then looking to the future, I guess what are SynSilico's priorities in the next year or two, would you say it's to kind of, you know, excuse me, raise awareness of your company and what you do?

13:27
Yeah.

13:27
So we want to bring data across to potential customers, have conversations with potential customers and then land a few pilot projects through those discussions to validate what we are doing and that this is delivering value for our customers.

13:50
So this is the short term goal that we want to have done within actually this year.

13:56
So this is more or less our short term target on board a few pilot customers.

14:04
Yeah.

14:05
Make your offerings very specific to the point that customers wanted to see.

14:11
And then, yeah, that's short term and then I guess final conversation.

14:18
And then what is the broad division in the long term over the next three to five years as this company, you know continues to expand and you know, hone in on the kind of like kind of specialised in the offerings when you have these nailed down.

14:37
After this initial onboarding of pilot customers, we want to of course scale up and that requires also growth of the organisation so that we can rapidly deliver the services that we want to deliver.

14:54
And we will scale up simply the number of projects that we can execute on for more customers, which brings us then into a growth and profitability scenario, which is more or less the next milestone.

15:07
Then during the next, I would say two to three years, this is what we are aiming at.

15:16
Sounds good to me.

15:17
And I guess final question from me, Oliver, is do you have any advice for people who are maybe considering working for a start up in drug discovery specifically with AI?

15:29
What would be a piece of advice you would give them?

15:35
I would say if you want to join a start up in the AI space, you want to understand their vision and their mission and whether what is what they are doing is realistic.

15:51
Because you don't want to join a company that is unrealistic in what they promise to do and yeah, and have not the capacity to do that.

16:02
So a good conversation with the team members and understanding their specific knowledge and what they are doing and why they are doing it is something that I, I would urgently, you know, ask people to do during their interview.

16:20
Because you're not only interviewed, but you should also during your interview, understand and you know what the company vision is.

16:28
And if they can deliver now, that's great.

16:33
Well, thank you very much, Oliver, for sharing your thoughts and some of the work you are doing at SynSilico.

16:41
It's been a pleasure to host you and we're very excited to see you know, where your company goes forward in the future.

16:47
And to our viewers, please do stay tuned because we will be hosting more insightful conversations with thought leaders in the drug discovery space.

16:56
Thank you very much.

16:57
Thank you, Lucia, and see you at the next conference.

17:00
Thank you very much.

17:02
Bye.