Gayblack Canadian Man

Foreign Policy Analysis
Does it pay off to organise your data?

Does it pay off to organise your data?

Thank you. Wow, incredible that the room
is so full! Hi my name is Hannes thank you very much for introducing me. I am
just I just flew in from Berlin yesterday and I’m very glad to be here
and stand in front of such an amazing crowd at this wonderful campus. So, indeed
I am the founder of digital entrepreneurship hub so I’m involved in science
entrepreneurship for a couple of years now, I coordinate entrepreneurship
education at my university for the last 7 years and currently build up a
science entrepreneurship program for all 3 Berlin universities so this is really
what drives me right now thinking about ways how to make more
postdocs go into entrepreneurship. I’m not sure whether this is my job but maybe that’s
part of it and so thank you very much for for introducing me in such a way. So
indeed I am involved for the last one and a half years on the question on how
startups use bio data so we have travelled across Europe particularly
interviewing founders of startups and to figure out what is behind their use of
bio data and I would like to share a couple of our insights with you and be
particularly open to the discussion that we might have afterwards so a couple of
issues that I may raise and also maybe the conflicts that I might bring up here.
Good, so the question overall is leveraging
infrastructure so the so we started with ELIXIR looking at okay how do these startups use the biodata infrastructures they can provide here for instance on
this campus so then I would like to start by the question of since I know
that this is a rather diverse group and not all of you are entrepreneurs but the
question of what is it that entrepreneurs actually do so what is it
job what it what are an entrepreneur a doing and actually when you go to an
entrepreneurship class I just went to about three days I was on several
entrepreneurship classes like me what you regularly hear is this kind of talk
so most startups fail because building a product for months or someone sometimes
years leads to failure because most startups
don’t understand their customer groups so what we do in research
basically so is we build platforms for instance for a while so I created a
platform for over two years just to find out that the customer segments that we
were looking for was a really bad customer segment and we tried the next
one and that failed as well and then I ended up in academia again for some
reason so this happens all the time so I see a lot of science and scientists
starting our companies with a very very limited understanding of customer groups
and this is what you see regularly also at the market so this is a very nice
example of one of one of these famous Silicon Valley startups that just
created a product that nobody really wanted to have it’s like a juice maker
that cost $700 I guess and gives you the packages that you bring in to it’s all
proprietary and so each juice costs cost more than tens of dollars so no one
really wanted to have that however they found some investors for whatever reason
in the end of startup failed because they were they just just creating a
solution that no one really wanted so long story short but entrepreneurs regularly doing is or what they have to do is they have to figure out what
problems do exist what kind of solution cannot provide and this is also a
digital entrepreneurs do so as Kathi already mentioned I found this digital
entrepreneurship hub as a digital entrepreneurship is a particular sub
theme in this kind of entrepreneurial entrepreneurship research. So we are
looking at entrepreneurs who use digital artifacts for platforms or
infrastructures to provide solutions to existent problems. So the good thing
is that there are a lot of different platforms and services out there that we
can write and use, integrate into each other and then create an interesting new
product to an existing problem and what we do essentially is we just link the
existing solutions that we know and lead them to problems that we find in
the market. It sounds pretty straightforward
and this is actually building on a lot of research that is already going on
since the 60s/70s for instance by Herbert A Simon who was very
much looking into this problem-solution mashup and
entrepreneurs do that because that’s like there’s a main job so they have to
find real work problems problems that are so dear that someone is willing to
pay something to get this problem fixed and then creating a solution for that
problem in the end you see all the different problems in there that are
potentially out there which we recurse referred to as maybe a proper landscape
or a problem space and we the entrepreneurs have to pick some of the problems
that they want to solve but then find the best solution or one solution that
is able to fix that. The gap between both is that on the one hand you have these
problems that might if I solve them might give a benefit on the other hand
you might have a solution that comes with a certain costs for instance if I
would want to use some scientific insights to solve the problem out there
it might come with a lot of costs right? So if I want to solve some kind of
cancer out there so let’s find it come up where they come up with the drug it
might involve billions of dollars to actually bring this bring this drug to
the market as you all know I guess so I’m taking this digital entrepreneurs
however are in a little bit different situation than many of the entrepreneurs
have in the last the the decades that happen that have come before so digital
entrepreneur leverage like the technologies that are out there in new
ways what does that essentially mean is that
the solution space is also out there so the many solutions there that are out
there in some way or another enable us to reduce the costs of bringing them
into the market. Let me put that into perspective. If you would want to start
in artificial intelligence startup 15 years ago you would have needed a
couple of computer scientists maybe maybe a mathematician right to
find a solution to create neural network for instance or some other
machine learning technique and pretended to market and saw some kind of a
real-world problem maybe reducing fraud. Today I’m currently
monitoring a team consisting of only business administration people starting
company in this kind of field using your own networks, using existing technologies,
particularly Google tensorflow, which just reduces the costs for them to
bring them to develop a product that is rightly marketable and this and this
venture has now just raised another round of money so they were they were
able to bring a product into market to face an existing problem
just because Google provided them the technology to do that and I think this
is one of the main leverages of digital technologies today we can use them and
easily prototype new solutions and digital entrepreneurship however doesn’t
stop there so it’s not about only creating a one-time solution so
therefore a little bit struggling with all the Entrepreneurship classes that
I regularly participate in it’s not about solving one problem once it’s about creating a solution that solves problems
at scale and therefore we learn that digital technologies again are very good
at this right so we can we can use a lot of different mechanisms come with
digital technologies to create scalable solutions and then there is hopefully an
investor who also believes on this kind of scalability and throws maybe tons and
tons of dollars on us say we become this kind of a thing what we now refer to as
unicorns. So digi-technologies in that sense are pretty nice because like they
separate the medium from the content as we know right so notice you don’t need
any don’t need any like printed out posters any longer or CDs just go into
the cloud and the digitalization has reduced the marginal costs of
our information for instance it also enabled us to build flexible products in
a very flexible way. For instance as I as we are having so many libraries
available we are now in a situation where you can actually take a group of
three people put them into a hackathon within a weekend and they can come up
with a solution and if I give them two days more they can come up with a
comparable solution but maybe on another medium so maybe it was first a web-based
software and then they just take similar libraries put them together in a
different way and then maybe it goes to Apple Apple iWatch so different
technologies gives it give us a lot of opportunities to build products on scale
so this is my basic line of research so this is what I’m really interested in
and as I said in the last year’s I’m very much interested in how do
scientists actually create solutions and become entrepreneurs so
the basic research question that we stated for ourselves is how do these
digital entrepreneurs match problem solution from science data? So not only from from like in in e-commerce spheres that
you might think of when you think of digital entrepreneurs but how
did they use science data in a way that is actually scalable so there was the
premise with which we started so we started looking at science data. So
I can’t you anything new here I think you’re much more familiar with
these kinds of things than I am so one of those papers that that I’m
referring to is actually originated here so you know the genomics data is like
growing exponentially there’s paper that I’m citing from is actually
saying that most of the genomic data that is going to be available in the
near future it’s not coming from the public but more from the private sector
so but anyways right now we have already a lot of genomic data available
for us so a much of this data is actually globally globally expanded so
globally disseminate different research infrastructures most
of them right now still on public infrastructures particularly in both
Cambridges so over here and the Boston area so it just has definitely some
historical reasons but also you are more familiar with better than I I guess. This
is great because there’s this data is available not only for
scientists but potentially also entrepreneurs so there could be startups
or there or startups out there but that might use this datasets. Data for us is
however nothing that we just say is facts so we acknowledge that as you as
scientists you create data you created with the particular intention and this
intention, and therefore I think is important, is maybe not to start a company
so when you create data or you manage it or organize data you so I guess you are
all scientists all most of you are you create this this data with a particular
particular problem mind which is regularly testing hypotheses or creating
a hypothesis. However if that is the case data has to be in has to be
interpreted depending on who’s created it with what kind of skills what’s the
background knowledge associate with this kind of data sets what are the
circumstances under which it was researched and therefore unavoidably
data is value later so you cannot take data out of its original context
just as easily as one typically might think when thinking about data so the
data is more than just zeros and ones and this is important because as I learned
regularly when I do these post doc workshops what scientists regularly do
so the kind of goals that you have and they can practice practices that you
perform fall significantly apart with what entrepreneurs do, So many of the
things that for instance scientists do is – regularly but not all of it of course – but creating datasets for for example for one time purposes so
like creating a data set to solve a particular particular problem a
scientific problem, so testing a hypothesis once and then once it’s for
its tested creating a publication or a bit hopefully get this publication
published, and paraphrasing here, hope they hopefully get this publication
published and then being happy about this and maybe a follow-up publication
can come later come on later. However this is not what entrepreneurs actually
are looking for watch news have to build anything that they create that they
create has to work repeatedly because they want to sell a product regularly in
an environment that regards very high has very high quality
standards therefore science data for from their perspective might be a common
pool resource however is messy as far as an incomplete oftentimes unstructured
and non-standardized and this is a problem for people who are working in an
environment that has this high quality standards for engineers with
pharmaceutical companies do so and the other thing is and this is something
that I learned also while spending time with scientists from from the biological
sciences it seems that scientific advance leads
to changes of metadata all the time it’s because you advance research
so which which which means that the way how data and science is
structured changes all the time and this kind of a change of the data structures
leads to problems when you want to apply a product that has to be as stable as
possible alright so you have to yes somehow these startups have to manage
the kind of validity that you create as you create new as you create new
research advances so I I’m laying this nameless ground I would like to share
with you these start ups are actually doing so right now in Europe. So I’ve looked across Europe and looked about what are the
startups that we currently have in Europe most of these startups are
actually in the United Kingdom many of them originate or somehow are
linked to to this place [the Wellcome Genome Campus] which makes completely sense when it comes to
bio-data companies. A bio-data company for me as a business is a company that
relies on a business model where data from the biological field made also be
life science is a key resource so it’s have any data that the data is in the
core of their business and this can mean many different things this can mean
either it is an app that does that provide some kind of healthy living such
as fitness genes where you I’m not saying that it’s good products I’m just
saying when it’s out there right so this goes to all of them Fitness genes where
you’re from okay you upload your your-your-your data and then they give
you a advice on how you should go on your fitness regiment or these health
reports that I mean I’m sure that you are aware of from 23andme where people
where they have where they have different business models on using
genomic information. There are also diagnostics tools are deeply
settled within the within the life science industries that are actually
used in practice and perfected by by practitioners in the medical field and
go nowadays even beyond from ancestry where you have these ancestry reports to
DNA Romance which is a very crude startup it tries to match people based on the
genome with many ethical concerns I would say. So as you see there are a lot
of different applications that you can see out there and as I want to show you
most of them if not all of them rely on somehow some form or another around the
research that has been in the genomics field. So we did a survey
on these startup founders like only get filled up by c-level founders of these
startups so far 43 from them and ask them a couple of questions among them “to
what extent do you rely on these different infrastructures that are
provided in the open research data” and what we found is first of all all of
them said to some way or another without any kind of research data that is
available we would not exist and interestingly many of them applied several resources being the one of the upper layers those the one that I
mentioned by most of them. So one of the one of the core findings for us was
really is it is for the startups most of the time very important to have these
research infrastructures such as ChEMBL or PubChem or and all of the other infrastructures having them available
because if they wouldn’t have them they would not be able to start a company in
the first place right so we didn’t stop there so that was for us kind of the
first step to really understanding whether they use this kind of science
data and then we wanted to figure out okay what are you actually doing the
science data I’m not going too deeply into into that but what we wanted to do
is we wanted to do a more qualitative study so mean that we that we
interviewed them that we went went into the field trying to understand what
these the startups are actually doing and created in a study of 15 startups
that we think are fairly representative of the startups that we see across
Europe and figured out what what they are actually doing with data with the
science data and we came out with three different mechanisms and wanted to
relate them to how they actually then scale their business and I would like to
share a couple of stories about them if I may. So one thing that I see
that I see to some form or another maybe all of those startups have done to some
extent or another so is what we refer to as contextualization this
contextualization is a process that takes science data and links links the
semantics very closely to a goal that is driven by a business problem so what
these companies do is actually so the first thing is they customize the kind
of goal so they figure out ok what goals are out there that can be solved and
then customize the semantics that to hit these goals let me illustrate that on an
example so this is how how this work might might look like over time so at
first they start with okay so I am i I have a science research project so
there’s one company that that was working on one of these
infrastructures across it within Europe and they figured out well maybe this is
also interesting for for a company in the pharmaceutical industry so what they
what they did is they started project with which is kind of listening so it’s
a singular problem solution to one single customer which is huge
pharmaceutical company and they they started with as internet interacting on
a project based at this time with with this with this pharmaceutical company
and they provided provided the software started building the software for this
particular company so they figured out this is something that is of interest to
them once but not really scaleable so what they began then is they started
another project in a sort of different so a different problem something that
they still were able to do because they they knew how to solve it and and start
to to address the different customer in the same industry but with a different
problem and again project doing another project and then going
again and doing another project. This is the kind of mechanism going project over project over project over project. It is fairly familiar to many scientists and I think
this is one reason why many science entrepreneur many scientists who start
these kind of startups are pursuing this because it is fairly applicable to what
they know already like finding grants going to a project doing the next
project keeping my my project team running so seeing my startup my startup
is actually only my research group in a different way and at some point or
another they hopefully you realize that well only going for project is not
really scalable it’s not really something that is sustaining me so maybe
I have to create a product and this particular company for instance and they
found a pharmaceutical company another large pharmaceutical company to
create a repeated problem solving so they created a software this software
was heavily contextualized to this one customer so was certified for this one
customer was audited by one customer they created an ontology that is
only usable by this one customer so however this customer
luckily is giving them enough money to keep on going so this is the kind of
contextualization i see that in like i think to some extent in all of the startups
that that we’ve seen but most of the startups actually have gone further than
that so because contextualization is like
very context-specific it has one problem one solution. Another
mechanism that was that some of them have used on top of that is what we
refer to as decontextualization which is a process that detaches the data from
its goals by predefined semantics so they find some kind of maybe an ontology
someone referred to it as an uber ontology so they found this uber
ontology that links different ontologies that you created in science
maybe and link them together and for this
they like for this one uber ontology and then they test our where does this
uber ontology apply and this is how this might look like so they started
creating this uber ontology and this is like the solid line in main means
that it’s a repeated problem with multiple customers so there are a couple
of puffs of sawdust out there they create distant uber ontologies that
link between between Campbell and PubMed and all these different infrastructures
and create great link between them to be actually to actually help the
pharmaceutical companies for the first time search these different
infrastructures are easily applicable way so in a good way with an easy user
experience and then they gave it mostly give it up for free so it’s mostly free
that you can search these different infrastructures and then they provide
additional projects based on what these different cost a company’s fine so the
pharmaceutical company uses their search and then comes to an interesting
potential insight that wants to link their proprietary data now to this open
data they go to the start and tell them why don’t we build up a project and then
they start building up a project so the problem in that sense is that they then
link the proprietary data to the all the open to this uber ontology that they
created the first place growing expanding their ontology and
then they gradually expand their ontology which makes it easier for them
the next time to link the next company and then link the next company and so
growing the answer regularly they they expand their ability to to to attract
new customers in a very easy and more cost-efficient way and this is really so
this is what we call MD contextualization and
this is Warren college is just one example of this but it’s a very
prominent one and then the final a mechanism that we and that we found
sought freedom for twist more recontextualization so once you have
this predefined semantics so if you done this a couple of times
potentially you find you go into this recontextualization as a process that
links data with goals by lining is predefined semantics let me illustrate
that on a different example so what they what what you do at some point you will
be able to like address the problem find a problem that is fairly scalable so
maybe I found a biomarker to the particular biomarker to to address a
particular kind of disease or to to maybe and maybe predict the relationship
between so that’s pretty clear relationship between between different
people maybe and then I can build out the data pipeline to process this kind
of data and become very very efficient in the process what these companies rely
on is that these kind of semantics keep being very very stable which means that
when science get involved when they when they and when they get in touch with
research and for such a science infrastructures but they very regularly
say is that well these these datasets are very very messy I cannot really work
with them because what they want is they want a very solid and stable
relationship they say ok what we’re doing here is very high quality high
quality service we can’t rely on this massive data so everything every time
they relate they work with science groups always with research groups or
research infrastructures they double-check they do they do a double
quality check – miss – to control the kind of impact
that this misalignment might provide so the design very specific solution
so maybe based on one particular biomarker but since they build up this
very nice infrastructure since they know how to build how to like stabilize a
particular particular semantic they are at some point or another able to address
a similar problem which is not that far away from the first one but it rested
addresses tubular problem and then slowly gradually scale scale as they
address more and more problems so what we essentially then found is that this
is actually the way that we see most of these companies kind of scale when we
talk about scaling mostly we think of Facebook or these kind of companies
right so scaling is something this is not true for these kind of startups
these dogs cannot find videos of uses because there are not millions of
pharmaceutical companies obviously however I can sell a product a couple of
times to pharmaceutical company if I solve a couple of problems for the small
surgical company so all of them they are b2b companies most of them are HP
companies even the b2c companies like 23andme they actually make their money
with b2b so if you want to address each of the companies in that sentence then
you need to find the different problems that you can offer to offer a new
solutions to them so either you use therefore find these new problems or you
offer multiple solutions for the same problem and this is the way that we
found that most of these startups actually scaled so to cut this a little
short all of them or most of them sorry this kind of context realization start
just doing what you do research there is a new problem okay
this is new problems in your hypothesis I have to test it then okay so I’m
building happen we said a research project in my startup solve the problem
find next problem by the next problem file at some point or another they
hopefully find that there are problems that there are similarities between the
problems so for instance there’s one startup that has done over 120 projects
within the last six years 120 projects I think there are 10 people and at some
point two years ago they recognized well one of the problems that we regularly
face is how do we annotate this data so they created a solution to build
annotations and this is first of all they did it only for themselves so they
create annotation software to help them to help them be more effective efficient
in their projects and now they they launch a project for Entei they know
they launch the product by annotating so they grow the ball about how coming back
from this context realization because what sexualization is really back you
want to go the schedule for the company because you can only scale up people how
many projects like it can you actually manage I’m not sure how efficient you
are but I mean managing project means manpower and scaling manpower is really
really complicated particular to the pirate limitations I mean where should
you find all of these bioinformaticians to hire it’s really really complicated
so at some point or another you want to find different models so therefore we
found most of the services actually scaled so actually we’re able to go into
series a series paid to build up more more employees up to 200 employees to go
first 2d context realization I meant recontextualization right away so that
you they were actually able to like stabilize the semantics that is
something science all the time changes but they wanted that they are able to
stabilize for for their company for their for the for the business customers
in order to solve their problems so my final slide before going to before
I would like to need to open this up to discussion as well as so we hope that so
what we are currently having me on this we hope to to introduce these mechanisms
to showcase how you can actually use the use science that I build up companies
from it companies that potentially scale because I think this is something that
the current that we oftentimes undermine as when you think about entrepreneurship
how do we actually build build build companies that scale beyond a couple of
problems and not only solving one and I hope that we also like provide a couple
of insights for our own research as we now find a different angle on going on
this problem solution strategies in the way I think also that we provide a
couple of implications potentially also for research infrastructure because
something that I recognize and maybe cut through a little bit is what many of
these startups are building on is that there is a gap between science and
industry and the fair movement is actually opposing this so widened right
so I’m not sure whether you whether you understand what I mean so if I build a
search engine that makes all these infrastructures available
then I am making searching infrastructures easy for industry and
this is what I might business model that builds upon however if you come now and
build up everything fair searchable a little bit track there is a smaller way
this is what they currently building on I’m not saying that it’s a bad idea I’m
just saying there is a comp oh there is a potential conflict and I’m pretty sure
that in the end of the end of the day as we see if all of them they build on
research infrastructure a fair infrastructures will in the end leads to
more interesting innovative ideas but right now there are a couple of
companies out there who would be who would might might face partners from it
okay so with this maybe there’s a bit bit of contracts a conflict here I would
like to open up the discussion because we have and I would be very very
interested to hear your experiences also with with met with the companies I know
there are a lot of companies here on campus and close by that youth-based
thank you very much for your attention so I was talking about the last stick we did that link to something so the
question I want to draw I think it links to what you just said so when you looked
at these companies how much of the business value were in this sort of
technology space ie in I mean essentially in the semantic
axes proprietary semantics all right the linking technology and how much it of it
is in the business model or in the business practice ie how do you go about
solving the problem for the customer was the luxuries in a way harder to copy and
I think that relates to your question of fair in the sense that well if you make
take that more hey whatever that means but more accessible for more people if
your value-add your innovation is around that sort of semantic axis then you’re
in trouble if you’ve run you ad is on the sort of business model business
process practice then you have less to fear from that yes you could give us
things from the companies you talked to where they sit in that space right so so
I’m going a little bit question whether you can really separate both okay
because right now we are assuming that we can like that we separate the goal so
the goal of our suited company is looking for a particular diagnostic
whether we can separate that from the semantic semantic structure of the data
set and what I’m seeing is that many of them
they engage heavily with these customers to build up to really understand what
their problem is and then a high really customizing products to the to this and
this also extends to how do you build a how do you organize the data so for
instance a they take they take these mobile ontology and then they go behind
the behind the walled garden of company and then and then they fly it
apply it there which means heavy linking and to it so therefore it might be very
interesting to see whether the kind of assumption that they may is that you
that you can actually separate both I think there is a tendency towards trying
to separate it because they need to standardize it hmm
however so far I’ve seen that they engage in a lot of a lot of practices to
actually you know write it down to the customers so from what I understand most of those
companies are very knowledge intensive so would you say that the majority of
the startups in data-driven Life Sciences is created by people like
directly coming out of universities from well most of the people that I’ve talked
to actually they come from they come from universities so therefore I think
this also explains why it is the math is heavily biased towards certain countries
so for instance why are many people like former companies in the United Kingdom
well because many of the research groups are here so people are trained on
genomic data sets in particular institutions
so therefore when they go out there to start the company they don’t go that far
away regularly and also it’s like very good for them to be very close to some
of these research infrastructures so for instance there as I said there was one
group that was that was started originally from research infrastructure
and then they didn’t move so much so far away because they wanted to be still
involved what is changing there what what what are they working on so
therefore this kind of interaction helped them I know and so so so one of
the founders at least is always like a scientist from in from a research group
so and then regularly they take other people inside and then it really depends
on which kind of Industry they are so friends is when they come from
pharmaceutical industry if they address in a surgical industry so one of the
punishments in suppose someone was before start the company in the United
States and that grew a large their therapeutic company and and and sold it
there so they take co-founders with other knowledge but obviously most of
the time their researchers started so how important is the location of
their customers to me is that possible part of where they are based or do they
tend to please I mean regularly this farmer said I’ve
stated that after that we’ve talked to they some originated from this places so
they didn’t move too far away actually so that seems to be the main reason why
they found it there so even the Swiss company so we have companies from
Switzerland in our sample someone except from the people that we talked to and
those were also people from Switzerland and that was the reason why they started
their companies there obviously also Switzerland it’s very nice it’s a nice
place you start a company and if I’m a suit again but the main reason that we
actually saw these from the people that we talk to those were the ones that yes
they were give a base there made their make their research career in a
proximity or like somehow with Bomberos the way
I like the 2×2 creator the loading this is a big fan of yes so so actually it’s like very pretty
interesting because they move a little bit and that’s the risk profile yeah so
yeah so so that’s the that’s regularly the field where they at some point
because I think this is why you’re just scaling them it’s just interesting I
mean there’s companies they are very good in solving a lot of different
problems but monogram bikini and so they engage this project they
build up a lot of different maybe ontology so make more plants or plants
and so they create a lot of different ways of predicting surgeries for
instance so then they did across the board so they’re very good at it but
they’ve never been able maybe they have never been able to to standardize it
once once they found it that’s that’s the point where they are then able to
look for new investors and then to grow so when you’re here you tend to be much
bigger so the biggest companies that we have like this those are rapidly scale
are here so here we have one company there’s a handful
but the employees and responded party six years ago so those really ones who
regularly rapidly scale those are the ones were who is scaling also quite
rapidly but not as much and those are the ones we have even one company here
that since 2005 and they have not grown so much just because they go from
project to project they could we find it so if he’s asking first profile
I would like look for it those particularly and look for those who are
able to find a repeatable niche it’s all been they have like these search engines
for instance so this could be one set of search engines or they have a workflow
software or sanitation software such as something like that but if they are able
to build something on top of that like find find one product that can do
something rapidity on top of this so if you have created this on top of the
super ontology and then build on top of this uber ontology a repeatable
application we repeatedly solves the problem to someone or to multiple people
then those would be the ones with the less risky diversity program does that
urge yeah so so one company for instance so
they provide so they created and this kind of an uber ontology or motor sounds
really awful but it’s so so they so they were able to link different ontology so
and what they are what they were able to do is and they build on top of that
first of all it is search engine to like differentiate between different terms
that are close by but not us but do not make me in the same things I would say
and then they they created an API on top of that that they can and that they can
sell to different to different an integrator to different services and
then they have like a software that they pay by usage for this particular API and
that is fairly scalable so most of the companies examples that
you’ve referenced probably fall into the kind of genomics and health data related
fields at once find some did you kind of get a feeling for what’s the next big
thing okay so that’s that’s the time if you
would recommend someone now which direction to go in which of the business
model station so yes I’d like I said I mean it’s I’m currently looking very
much of these kind of search engines and there are at least two or three so there
are a couple of them in Europe already and these search engines I’m just I’m
very curious whether they what are the additional services that they offer on
top and there are a couple of very interesting ones but the future I’m very hard for me to
say so I don’t think that it’s so Peter C has shown to be not worth it so
because that has too much and and obviously the pharmaceutical so
pharmaceutical industry is obviously very very interesting and it has to be
seen that there are a lot of interesting to your critical usage for genomic data
but that’s because I think like your browser examples or leveraging effect
that we have well there’s none non-interoperable yes it’s and so dying
will solve those problems so but i think there is this step ahead I mean that’s
it I mean that’s hopefully their Savior I mean they are now a little bit of step
ahead because they integrate is different in their infrastructure but
now and now they have these integrated integrated data base and now they can
think of interesting solutions and they try them out with different industry
players and if they are able to find a solution on top of this they are the
step 1/2 in front of others so I think this is why I think that is are
interesting so
and they are mostly for both for public data is a free thing like all of them
they have this premium package so free is public data and and for proprietary
data they offer they have a premium pact they all have some kind of a premium
package so then they integrate the public data infrastructures with the
private infrastructures that maybe a pharmaceutical company or biotech
company has so and that’s why it’s an interesting for oh yeah some extent to work to what
they’re doing is some kind of affair right because it’s fun ball more or less
success yes I think the main differences and two things so first of all right now
even these search engines I mean if they provide a search engine for free to the
broader public and you can search the public infrastructure so this is kind of however so they own the way they created
this interoperability so they own this ontology and you don’t know how they did
it and you don’t know the process how they select what they show to you so and
this is in sound extent is good because sometimes this raises the usability to
an industry providers because industry doesn’t doesn’t want to see everything
they want to see only the relevant thing they want to see the Google results ten
results and that’s it so they want to see the relevant from the facts first so
this is in essence good on the other hand like if you would like to see all
the data that’s not very good and for for a company is very important
to have this kind of a black box because that helps them leverage leverage this
kind of data said later on doesn’t it one more question happy graduation I did coffee Constitution
so with licenses they like that’s that’s a very interesting
question let’s say so I’m not saying anything about the kind of but it’s a
really strategic approach I don’t think that’s a strategic approach I think it’s
it’s some kind of annuity that you that they do these kind of projects not all
of them of course so I know that some of them are fairly open about their strategy because they are just happy
just and they even say to me we’re happy with 10 people and this is we don’t want
to grow we’re happy doing this we have this one product customer very good but
yeah so I would agree to you to some extent that site it’s just what
scientists have done so far project project project but as you have a
company you have to find a sustainable model and particularly with when you
when there are external factors that could lead to you rethinking this and
this external factors are nearly always so much learning that goes on including
that Kim is not so sure but the question is whether they learn whether they
really downside accounted a couple that learned and like I said so after 120
projects we learn annotation software something that we repeatedly do so maybe
that’s something that we should do but not all of them are like we like you
have you really learning business just the next project comes in that’s fine so
like building a software for four years for one customer then getting it audited
for the next two years so like for work it for one customer for six years just
to get the software ready and then thinking that’s a good thing I’m not
really sure that’s the same approach yes but again
even software usually it’s in three months or six months oh no they really
work like yes yes I believe that’s what it’s yes but I was making analogy in
terms of it this is the typical South biscaia that’s true yes like an app I
can create an app within about this room

Leave a Reply

Your email address will not be published. Required fields are marked *