New Apps For AI in mRNA With Anima Biotech's Yochi Slonim

There you'll find hundreds of videos of my interviews with biotech builders , categorized by topic , like finance and capital markets , regulatory discovery and manufacturing . Don't try it if you listen while driving , but be sure to check it out when you get where you're going .

Go to bioprocessonlinecom , hit the listen and watch tab and choose business of biotech in the dropdown . Yochi Slonim made a name for himself in global tech hubs around the world , from San Francisco's Silicon Valley to Tel Aviv's Silicon Wadi .

Notably , he was co-founder of billion-dollar Mercury Interactive , which HP acquired for $4.5 billion , and , previous to that , a leader at Technomatics , which sold to UGS and eventually was acquired by Siemens . Those were the bigger deals that came on the heels of several other of Yoki Slonim's startups . What's all that got to do with the business of biotech ?

I'm Matt Pillar . This is the business of biotech . On today's episode , we are visiting with Yochi to learn about his foray into biopharmaceuticals with Anima Biotech , a company he's been building for the past 10 years .

Anima's approach to tech-enabled drug discovery and development is multi-pronged , yielding , as you may have guessed , a target and drug discovery and development platform , a handful of partnerships with some very heavy hitters in big bio and a deep development pipeline of its own . Yoki , welcome to the show . Thank you for having me . It is my pleasure to have you .

We're honored , and before we get into the anima story and go too deep into the technologies that you're building , we need to kind of get a background on you Because , as I mentioned in that intro before , you were a biotech entrepreneur . You were decidedly a tech entrepreneur , primarily in software . What motivated the move to get into biotech ?

And indeed I built and took public or sold several software companies and the way that this got started , I was working with startup companies as an investor and I was introduced to the technology idea of Anima by my co-founder and I became the first investor in the company .

I looked at this and he was kind of presenting it to me in a way that appealed to me because my last software company was something that actually I did . I did several software companies in the field of automated software testing and debugging and this company was actually quite easy to explain .

It was like a black box that you have on the airplane that records the flight when it crashes . You can see what happens . They dig the black box and they investigate and you can hear what happened in the cockpit and see all the different engine conditions , and so this was actually a black box for software application .

It would record everything that happens in the software application and then , if something happens wrong , you could go back in time and investigate what happened . So we were selling this to banks , financial institutions , e-commerce sites , all these complex stuff and I sold this company to BMC Software back in 2006 .

And I was looking for other stuff to do and then he came to me and he said you know , I have a black box for cells , for biology . It records everything that is happening in the cells and actually it records it in images and you can see the ribosomes making the proteins . And the proteins are made from mRNA , which I knew nothing about back then .

Yeah , and this is like a processor that executes software code and the mRNA is the software code and the ribosomes run over it and you can see all that in images and this recording is like a black box . Now you can go back and understand why a disease is happening , for example , because the assembly of the protein went wrong , the translation went wrong .

This kind of looked to me like a tech company and I asked him so we are going to build like an imaging system over all these sensors ? Yes , we are going to build it , and I had not much , you know , other than a hunch about it .

But when I decided to bet on it , I don't even remember why , and I became an investor in the company and this dragged me on kind of down a rabbit hole that was completely unexpected and that became a very long rabbit hole because the timelines I wasn't aware of , the timelines of actually doing things .

But then I became aware that this technology generates images and tons of them , millions of them , and this is like big , deep data . And it kind of attracted me that with all my skills in technology we could build a platform that would kind of look at all the data and start to analyze it .

And this was still an academic phase , but it drew me back very , very strongly into biology , which was a challenge by itself by the way , real quick , yogi .

It was sitting in the academia and the platform technology , so to speak , was developed .

So there was no like AI , but people were talking already about the early early , I would say phases of what became machine learning early , I would say phases of what became machine learning , and they were trying it both algorithmically but both like let's do it non-algorithmically , it's kind of seeing the conditions and just acting on them , based on patterns .

So that pattern recognition is the beginning of , if you want to say , ai in many ways , clustering of information , learning from big data . But it wasn't yet AI . But the roots of that which actually appealed to me were already , as you were saying , back many years ago .

Whenever they were not doing it , it wasn't . So you could see where , when and how much of a particular mRNA , specific one , is being translated . And I was looking at that and I don't know why , by the way , because I didn't know much about biology . We'll talk about it later how I became a biologist down the road .

But I actually looked at that and I said but this is valuable as an academic tool If we want to do something with it , what it's going to be , and I don't know why , because I didn't even knew back then the words drug discovery .

But I said to myself in every industry you need to go for the main street , not for the side streets , to look for the interest and the money . And so I said the interest and the money . And so I said maybe this could help us , you know , create diseases . And then my co-founder , a biologist and she said well , this is called drug discovery screening .

This is what we can do with that . We could actually turn this into a big platform that is looking at millions of these images and where there is light , there is protein being made . We'll be looking for small molecules that are modulating the light . If they turn off the light , they kind of inhibit the production of the protein .

If they increase the light , they increase the translation of the protein . And I was asking her I mean , this is something that people are doing . She said , yes , screening , but not like this . This is high-content screening at high throughput . We'll need to do deep , deep , deep screening .

Seeing all these images and doing this over millions and millions of them , over hundreds of thousands of compounds , and I don't know how to do that . I said , wait , this level of automation is like tech company . I know how to do that , let's do that together is like tech company . I know how to do that , let's do that together .

You know it's like the biology essays , seeing the biology with the automation at scale over the big data that is generated . This seemed to me like a tech bio , as it is called today . But kind of , we were a perfect fit , you know , for starting an idea like that , yeah , you mentioned .

But what were some of the maybe the more challenging aspects for you in your transition into biotech ? Learning biology , becoming a biologist , as you say ?

But actually I became very interested because I mean , I was as an investor , as I said , in the company for a while and then I became the chairman of the company and then I became kind of an acting CEO of the company while it's still , you know , in process , and I kind of started my journey and I decided that I'm going to do it lightweight initially and I

was taking courses in biology online universities and it kind of became five courses per year and now we are talking about several years . So all of a sudden I was like 25 courses in biology behind me at the time that we are starting the company . So I didn't know about how actually it is done in the industry .

So a drug screening platform I wasn't aware of that , but I had the tools already to understand what it could do . And you know the idea of pathways in mRNA biology , what they mean , that we could see them , we could visualize them , we could find compounds that modulate them . Already I had that in mind .

So now we've connected back to the tech side and I could envision how a large scale I would say massively parallel automated system could run in a lab and actually do that training and basically analyze all that data . So it kind of was at the right time , you know , coming to me .

While I understood already enough biology , I continued to study biology over the years and actually I can interact , I mean , with any biologists from a pharma company and I'm doing this routinely , although I usually deal more with the business side right now and the strategy of the company .

this is more than cool technology , we've got something that we can develop into a sellable product ?

And then two , I guess on the back half of that question , building teams , where building tech bio teams especially then in the lead up to the launch of the company , it was a new concept in hiring to outfit those teams with talent to this day is a challenge A lot of companies struggle with sort of the whole brain tech bio kind of approach .

So I'm curious about how you built the team . So , like I said , two part question . There might be a big jump in between the two questions . But one when was the aha moment ? We've got something that we can build a company around . And then two how did you go about building the company from an HR standpoint with talent that could meet your needs ?

So it gave like more and more and more , you know , kind of feeling that there is something really here and we should be doing a real company around the technology . It's so unique .

So I decided to take the technology out of the university after I saw so many different applications and users , so to speak , that basically had a very limited research license to the very early phase to conduct a single experiment , so to speak , and visualize their proteins and mRNAs of interest , and so that kind of occurred to me .

Naturally I wasn't at a point and say whether it is a company or not . It already kind of transitioned in my mind to what are we going to do with that as a company ? And in that meeting where I was kind of introduced and I asked the question can we do treatments of diseases , discovery of drugs ?

And then I heard that there is this concept of screening and big pharma companies are doing that , but not in that way and not with that technology . It was the aha moment for me Okay , this is big , if we could do this , this .

And then my question was right away , as you were saying about the team , because I built companies in the past and I know that this is extremely critical . There are many good ideas that people come to you with and they say I want to do this company and you think about it .

It's a great idea , but not for this team , because they cannot get it off the ground . They don't have the skills . So I was thinking this is biology , imaging , bioinformatics , a lot of software , big data , data analysis and chemistry . Eventually , I was not aware of that in the beginning , but we built in-house a lot of chemistry .

I was initially thinking this we can take from CROs outside and so on , this we can take from CROs outside and so on . So essentially , this is a multidisciplinary team and a hard team to build . So we had actually the biology and the image processing covered between the biologist co-founder and an image processing person .

And then I brought all the software people from my previous companies quite easily actually .

I don't want to say sales like in a salesy kind of way , but sort of that value proposition you go to the industry with and you say you know AI , our application of AI that gives you valuable , like actionable imagery is better than AI that spits out prose , for instance pros , for instance .

Okay , so that's actually that's actually an interesting questions , because we are seeing today ai coming into two different worlds the world of words and the world of images and they actually kind of came onto the scene almost at the same time with generative AI for language and generative AI for images .

And actually , when you look at it , you could say that so far , and maybe I can explain , I think , why AI is actually better in images than words , and the reason that this is the case is that you know , actually humans are like that as well . If you think about it , words are ambiguous . Images are just data . It is what you see .

Okay , then you come to explain it , then you need the words , then different people will describe the same image in very different words . So , basically , if you go and train AI on words , it gets all that ambiguity into it . So you train it on a million words and it has millions of ambiguities .

If you train it on millions of images , you are just giving it millions of examples of the thing . So , actually , if you will actually look at this in the context of applications , for example . Actually , look at this in the context of applications , for example .

If you try to explain what is the difference between a dog and a cat and put it in language as rules , you will run out of paper space before you figure out that for every rule there is an exception and you cannot actually say what is the difference . But when you see a dog or a cat , you know the difference . You know if it's a cat or a dog .

Now AI will find it very hard to explain to you the difference between a cat and a dog , but give it a million examples of a cat and a dog to a neural network that is seeing those images . Give it now the million and 10 image and ask is it a cat or the dog ? It will immediately classify it correctly . So this is a very robust application of AI .

The most robust application of AI actually is the ability to see in an image something and classify it versus A or B . Now why is that even relevant to all we're talking about ?

It's about the ability to make sense of a lot of data , and this is what AI is really good at . What people are not good at is when it to make sense of data . We are good at , and AI is actually sometimes struggling to come up with reasoning in quite simple situations , but when it's a lot of data and you try to make sense of it , you can't .

So biotech is all about biology , and biology is a lot of data , and you try to make sense of it , you can't . So biotech is all about biology , and biology is a lot of data , and it's unrelated data that people are trying to connect together .

So there are millions of publications about diseases and proteins and roles and targets , and molecules and drugs , and it's extremely challenging just to understand the links between all of these elements and you are biased towards what you know and what you have read . You don't know what you haven't read .

So you can now use AI to summarize those things for you and find connections and discover mechanisms by kind of connecting that if A is related to B and B affects C and C is a protein that is doing something with D , then there is a connection between A and D , and it's good at doing that and it summarizes this into knowledge graphs that you can visually see

it and this is actually a powerful way to look and learn about diseases and mechanisms and targets that could be involved in diseases . But notice one thing okay and this comes again to the words versus the images . These are words .

It is good in connecting the words when there are many , many words , but it is starting with the ambiguity that is in the words themselves . These publications have ambiguity . The data is incorrect , the assumptions are incorrect . It didn't understand what is being said .

Really , eventually , the conclusions are not what is said in the abstract , but the abstract is telling the article . So that's kind of forward-looking statements into what could be done . But the experiment didn't prove it and everybody is citing it back then afterwards as something of effect .

So AI actually is good in images and images are a completely different way to approach the whole problem . So if you're looking at experiments all these publications , by the way , are post-after-the-fact summaries of priorly done experiments , right , but those are very biased because of the words and because of the connections .

So each of them may be good but many of them are not , and then the connections are wrong . Now imagine that you could do it in a completely different way , and this comes actually later to what we have done also at Hanuman , actually later to what we have done also at HANIM .

If you're doing an experiment and you image it in images from inside the cells , looking at the biology at play , you are basically observing the real biology . And most experiments really those in publications , for example they are measuring one parameter .

But if you could actually visualize an image , just like we people visualize a cat or a dog , we visualize at once a thousand parameters , a hundred thousand parameters . We cannot then say what the parameters are . We don't think like that , we just visualize it . So imagine that you could generate millions of visualizations . For example , a disease like Alzheimer's .

You take and you visualize 100 million cells in parallel at once , automatically , cells that are diseased and cells that are healthy , and now you give it to a neural network with AI to look at that and say what is the difference ? Now , it will not tell you the difference in words , but it will understand the difference , just like the dog and the cat .

It doesn't explain what are the rules , but it knows now . If they see a cell , is it diseased or healthy ? Now , this is huge because now if you can actually have the technology that is seeing , as we are seeing , the real biology pathways , for example , in mRNA biology now you could go back to that knowledge graph , so to speak .

Now you're asking a much better question . You're telling him the difference in Alzheimer's disease between healthy and diseased cells , based on 100 million examples that I've seen , is this pathway . Go and figure out for me now , with your abilities , what are all the proteins along this pathway ? Is any of them known to be associated with Alzheimer's ?

And now we are really into understanding the disease mechanism . But AI is more helpful , actually , if you give it real data , unbiased data , 100 million examples of it . This is where the real power of's , I guess , perspective at you .

So this is where I'm going to ask you for some differentiation , because the perception among perhaps many of our listeners who have only had that kind of consumer grade interaction with Chat , gpt or whatever other image-based AI tools are out there on Google , is like , I'm not so sure about the accuracy of the data , because when I ask it to create a picture of

a woman sitting on a chair , oftentimes the chair has five legs and the woman has six fingers on one hand . So we look at that and we go . Well , I don't know if AI is that good at imagery .

What's the difference here between the applications that you guys are building and they ensure , I guess , your assuredness around the accuracy of the tool in your application versus you know what so many see from sort of a consumer level ?

I am actually talking about image understanding by AI , given an image that is the real image . It's not an image that AI generated . For example , tesla is already offering self-driving vehicles . How does that work ? And this is working .

This is the most robust application actually of AI out there and it has been working for a while , because it comes back to something that the AI is really really good at , because it comes back to something that the AI is really really good at .

Tesla has cameras that were on the cars driving for a couple of years , on 10,000 vehicles and taking videos , which are broken into images of what roads and driving looks like . This is a stoplight . This is a road . This is a car . This is a person . This is a tree . This is a sign . This is a person , this is a tree . This is a sign .

This is that you are seeing millions of examples of that , and then it becomes so much information and so many examples that it can actually drive in real time and knowing what's around . I'm talking about something that is extremely simpler than that , even , which is let's look at 150 million images taken from cells of a disease model .

Now , these are real images that indeed we take with our imaging technology , but they are the ability to see and visualize the actual biology . It's not generated by AI , it's just taken with a microscope . It's a high-powered microscope that runs in an automated lab and takes 150 million images from all of these single cells . So the data is there .

This is the biology . Now we apply image classification and understanding on top of it , just like the cat and the dog . Okay , it's a completely different paradigm . Yeah , okay .

There's really , no , not even like a sort of a time parameter that skews the data , because it's just , it's real , it's potentially real , potentially real time , I would assume . I mean the time factor is going to take it off the table .

What does the translation of the machine's understanding of what it's seeing to the human being who can do something about it look like ? Does it remain visual through the application state , or is it at some point ? You know , is it subject to analysis in a visual form and then decisions are made based on that visual analysis ?

I guess I'm wondering what the action point is at the end of the process .

So , really , in a nutshell , the our , our platform it's called mRNA Lightning , and mRNA Lightning Platform visualizes and decodes the mRNA lifecycle and is used in multiple applications of drugs and target discovery RNA drug optimization , mrna vaccines optimization anything that relates essentially to , I would say , drugs and vaccines in mRNA .

Okay , which is a particular specific area in biology , but an extremely promising one and a very big one . Okay , as we've seen with COVID , for example . Yeah , so we have really an automated mRNA visualization screening technology . We generated over the last decade , more than 2 billion images of mRNA biology .

This was done in dozens of diseases , cell types , conditions , and we use this massive data set to train disease-specific what we call AI imaging neural networks , and they are capable of recognizing what we are calling disease signatures .

This is really the difference between diseased and healthy cells in a disease , and it's a particular mRNA biology pathway , because the visualization technology enables us to take 50 different visualizations in each cell . Each of them is a pathway in mRNA biology regulation , so it's something super important in the regulation of the mRNA .

Now , what happens now is that once the differentiation between healthy state and disease state is kind of seen in 150 million images , the AI neural network kind of extracts the disease signature , that pathway , out of it .

It tells you this is the difference , and this is actually a huge step towards answering the biggest question of all , which is why is the disease happening in the first place ? What's the disease mechanism ? What's the underlying disease mechanism ?

Of that deep data in images , you have an image bank that is almost like the disease atlas , if you want to call it like that . It shows you why the disease is different from healthy state . However , you want to actually do something about it . So this is where the error analysis of the massive data set actually comes into play and turns it into numbers .

It quantifies everything for you and it analyzes it and it suggests novel targets and it enables you to get started with the next phase , which is screening for compounds that would modulate exactly that pathway and will identify for you a drug basically .

So it's a drug and target discovery platform In the same way and , by the way , this data that is generated and gives you like unparalleled insight into the disease mechanisms and it aids you in optimization of drugs and discovery and prioritization of targets . The words are coming now back .

The words world is coming back because into the system we plugged in a large language model , which is the mRNA Lightning LLM . Now we are not in the business of developing LLMs Nobody is other than a few companies in the world but you could bring them in if you have data that is robust , that is unambiguous . They will be actually very good at that .

So we are bringing in LLM and we've built a chat-based interface through chat GPT , which we call the MRNL Lightning Copilot of the system , and you can interact with the data in natural language . And all these analyses , screens and all the information is actually presented by the system .

So it's an automated screening now of compounds follows and we identify hits , compounds that modulate this and then we optimize them . So it's a full-scale drug discovery and optimization platform .

More recently we developed applications that are for vaccine optimization or RNAI drugs , sirna ASO optimization , and all this platform really runs in a massively parallel automated lab architecture and the screening is really high content , but it's done in high throughput , like huge data that is generated and it's finally also an open system , so you can extract all the

data to other AI analysis tools in case that our analysis , for whatever reason , is not what you want analysis , for whatever reason , is not what you want .

And you start out you're like , hey , we could use this , this is a great tool for target discovery or a great tool for molecular discovery , small molecule drug discovery .

Did the additional applications kind of unfold as you saw , the technology having a fit there , or are those things that you thought well , we need to go chase that down because mRNA vaccines are a big market opportunity and we need to make our application work there . How did that evolution , I guess , of the platform happen ?

So it's so much harder to succeed in a company where there is no market for what you're doing than to succeed with a product when there is a market , even if your product is not so good . And our product excels . And we started it in small molecule drug discovery , that screening of compounds that basically are modifying the image as they are in the cells .

You kind of turn it into a massive , massive , unbiased drug discovery training campaign by taking , let's say , 300,000 molecules and each of these , when they are in the cells , if they do something to the mRNA biology , they will change the image . And now we are back to the dogs and the cats .

But now we are looking for a molecule that will make the image look more like a dog and less like a cat . So the hits are those that are modifying the images , from the diseased to the healthy , and the AI has a neural network that can tell , given an image , what it looks like . More so it flags for you those that are active on the mRNA biology pathway .

So you know that you've got a compound that is active . You even know already almost where it is working , and this is huge . Maybe we'll get to that later . About the mechanism of action , usually that's an after-the-fact fact , hard and mostly not successful , unsuccessful challenge . Here you can get it even before you start . It's just a guide into it .

So when I looked at this I said , okay , we've got this working . We did three partnerships with big pharma companies Lilly , takeda and AbbVie . We've got that application fully figured out and we are developing a pipeline of our own .

And I was around presenting the story , talking about it , and all of a sudden I start to hear people coming to me and they say well , we are in mRNA vaccines and if we could visualize the vaccine which is an mRNA inside the cells , with your technology we could actually profile 200,000 different designs and you can actually visualize which one looks better .

And I ask them what does it mean look better ? In my terminology , healthy and diseased , you know I'm moving from diseased to healthy . But here is a vaccine and they're all the same cells . They say no , healthy and diseased , switch them for less efficacy , more efficacy .

So we want the vaccines that are translating the most protein and in the right place , and we want so . Could you actually find ? You know for us , by comparing all of them , which ones look the best ? So let's throw 200,000 into your system , massively parallel environment like this .

You screen them , you come back with 10,000 , with 100,000 , or 50,000 , 40,000 , whatever the screening success rate is . It will usually be like 20 to one reduction right away . And then you throw them back and you say , okay , permutate a little bit about those , bring them back for a second optimization run , and now you will have maybe only 2,000 .

And every time we visualize and compare and it's a race between the vaccines you can see that this is the same thing . Actually it's the same technology applied to a different problem . Now siRNA drugs have this problem , which is the endosomal escape challenge .

99% of the siRNA actually gets trapped in there and never gets into the cytoplasm of the RNA of the siRNA actually gets trapped in there and never gets into the cytoplasm of the cell . But we see it , we see where it is . So we could do things you know that modify and enable you to choose those that are actually best in doing that .

So again , it's the same stuff . So if you have a machine like that , a machine it's not a machine , it's an architecture , a platform like that and it can visualize the biology inside cells . It's applicable to all these problems .

So now we are interacting as potential partners with companies that are coming from vaccines and siRNA drugs and we are exploring the value of the technology in partnership mode . We are not going to develop those , but we are going to offer the technology for those problems .

We know it's safe , we know it's effective . Name your drug . So why is it important now ? Give us some perspective on why this advantage of your platform that gives you early access to an understanding of mechanism of action , why is that important now , when we've kind of been living blissfully ignorant of so many mechanisms of action for as long as we have ?

If you speak with a pharmaceutical company and you've got a compound , and it's a great compound , but you don't know how it's working , then unless you go all the way to clinical trials and validate that it's safe and the efficacy is there , they will not want to touch it .

So , and the reason is that the side effects of that compound could be that it touches upon targets that are bad , targets that have failed already in the clinic .

So the idea is this yes , if you have the money and you want to go by yourself all the way to do all the clinical trials , and only there you will find out , or maybe you don't have to find out . It's just a matter of spending so much money and maybe you could avoid it if you knew the mechanism of action .

Because if the mechanism of action is going through certain proteins , kinase for example , that has multiple functions , and your drug is hitting that kinase as a side effect that you are not aware of , that's like already known to be downstream big problem and they will avoid the project altogether .

So I've seen that pharmas are actually very , very sensitive to know the mechanism of action and the target before they will actually invest , to go to the clinic with the product .

You can develop them all the way to preclinical and not know it , but it is hard today to convince people to go for , I would say , a project without the mechanism of action and it forces you to go into the clinic almost on your own . This is a very expensive proposition , especially if you have a broad pipeline .

However , if you look at what is happening today with technology , that's a very , very challenging problem . Given a molecule , how does it actually do its magic ? It's a harder problem than fighting the molecule and how do you solve that ?

So we used to solve it by a technology that we introduced three years ago called MOAI Mechanism of Action with AI , which was kind of taking a backward look . Here's the molecule , here's the data that we had before during the screening . Try to figure that out for me . But that's really challenging still , even for us .

I mean , we had great success with it and I would say five out of eight projects in the last eight , I would say , in the last three years , we've done it successfully . That's 10 times better than the success rate of the industry already .

But what we are doing now with the neural networks and the ability to actually see right from the beginning what is the pathway that is different between the healthy and the diseased cells . It's like already knowing the MOA . Almost you are there in the neighborhood of the target . So that's a game changer . That's turning the pyramid on its head .

It's not starting the project and hoping to find the MOA in the beginning . It's starting the project , understanding what you are going to target and then actually screening from there . So it's really an interesting proposition .

Now you can do this , as some companies are trying , with words , as we said , looking at the publications , looking at the disease , what is known , come up with ideas for targets . I think that our approach , which is unbiasedly just looking at the biology , give the AI a hundred million examples , Just looking at the biology , give the AI 100 million examples .

It will tell you the pathway . Now go with that pathway to the NLM and ask it what it knows from publications about it . It's a much more powerful approach in our opinion .

You know optimization , but at the same time you're developing drugs , and you know , the more I talk to you , the more I wonder why , like you've got this platform that does amazing things , you've got partnerships with Big Bio who are leveraging the platform to enable their own discovery and understanding efforts .

You know you've had great partnerships with academia on the technology itself . Why not just go to market with the technology itself , offer it as a service and let other people do the drug development ? Why did you build a pipeline or why are you building , I should say , a pipeline , yeah .

Has anybody else done that ? So either you show that you're doing it or you show another partner that has done it , or several of them . So we had to develop a pipeline , first , to prove that it works to ourselves . Second , to prove to partners that it works . Third , because there is value in those molecules and we were able to fund the development .

You know , it's kind of feeding each other . You develop your own pipeline , you use your platform , you understand how your partners would be using it as well . You fix things , you improve things , you test it . So you're using your pipeline as the way to advance your platform and you use your platform as the way to advance your pipeline .

But if you ask me what the next step is , that's the inflection point that you need to take a decision , Because if you want to go to the clinic and we are pretty close already there okay , so then it becomes a decision how do you continue to become both ? And this is a decision that we are thinking about .

We are actually , you know , by the end of maybe of this year or early next year , we're there and it requires a different you know , maybe thinking about it , but we are seeing so much interest in the platform right now . Ai for mRNA biology is unique .

We are , by the way , the only company we've been in a peer group of 32 companies in small molecule mRNA drugs . Go and find another one that has an AI story to tell .

So we were , from the very beginning , very different , but we were in a place where we were looking similar and now we kind of diverged from that peer group altogether and we are really showing why the platform is so different .

Because it's really an AI tech bio platform on the basis of that rich , deep data that is generated from millions of images , completely different from the target-based approaches of designing molecules to bind into mRNAs or affect its biology . In a certain way , it's a different idea , different concept , and now we are seeing that it's much bigger than we thought .

It's not even just limited to that . You know , mRNA , small molecule drugs . It's so much bigger than that . So all this strategic thinking is actually happening and is driven by a pull from the market , less than a push , you know , by us , and it's a good place to be and actually I like this kind of interaction .

probably partner out ? I mean , maybe it's too early to tell . I'm not asking you to . You don't have to answer the question if you're not prepared to or able to .

There is one company that did actually five , and then the next two companies did two , about five did one and all the rest never done a single deal .

So it tells you that the field is very much in the beginning , in the early phases , and the farmers are still looking at it , learning and trying , but it's not like completely all over the place and everybody's looking for it At the same time . If you think about it , mrna is in the market vaccines and in RNAi drugs .

So to apply our platform to things that are close to the market or in the market , from a business standpoint it makes a lot of sense , but I don't want to develop them . I want to work with my partners there .

So I definitely see Anima continuing to do partnerships , and many of them in many different areas , and we are , at the same time , moving our pipeline forward .

It will come to the point of whether we license it out , whether we are continuing with certain ones , maybe with spin-off companies , like some other companies have done , to take individual assets to the clinic . That's also a possibility . There are many ways to go about it .

But you know , if you care to share any more detail on the partner , you know , I guess , the details behind the partnerships with those three companies .

But for me what is important is really not all of that , because that's just an evidence that someone else is seeing value in your big vision and in your technology approach . Otherwise you don't get these partners and definitely not at that scale . And I think that there are two things that they're looking for tactical , which is molecules .

They're buying mostly molecules and in the environment right now it's all like shopping for molecules . Molecules are cheap . Pharmaceutical companies are loaded with cash . Biotech companies have problems to raise funding because of lower valuation , so they sell assets and it's kind of a lot of assets for cheaper on the streets .

So to do a technology platform deal it's even harder in an environment like this unless you are really really valuable , because that's like investing at long-term strategic stuff when you can buy with your money something that you can take to the clinic right away .

I think that we have that and I think that our deals were like that and because of that , because this is strategic , this is transformational AI with mRNA biology . I think that's big , so I'm very motivated to continue to do that and we are getting all the right signals that there will be more of that nature .

We are having so many different conversations right now about that , and many of these are with people that are AI people in those companies . We even have conversations with tech companies big tech companies like NVIDIA is a big tech company .

I'm not saying specifically we're talking to apply AI over data that is proprietary , is data that comes from diseases that have never been visualized before . Here is a visualization of a healthy and diseased , and what's the difference ? Bring AI with all its power and work on that . So we are thinking about all of these things .

This is very exciting , very big and very new and , at the same time , we are very , very connected to the reality of running , you know , a tight ship . This is a business that has to continue and has to be careful , because times are actually not amazing still , you know , to say the least , for biotech companies . So we are well-funded .

We have the ability , with our partners , to move forward . We've got lot of stuff going on in development and this is remaining our focus going forward .

I've abused it , but I'm glad I did , because I've enjoyed the conversation and I've learned a whole lot .

While you're there , subscribe to our Business of Biotech newsletter at bioprocessonlinecom . Backslash B-O-B . In the meantime , we'll see you next week and thanks for listening .