Latest / The BioHub - by Avetix / Episode 128 - Emily Gransky - VP, Talent @ Formation Bio

Transcript

Episode 128 - Emily Gransky - VP, Talent @ Formation Bio · The BioHub - by Avetix

• 0:06
  On this episode of the pod, we're joined by Emily Gunsky, VP
• 0:09
  of Talent at Formation Bio Formation Bio for if you haven't
• 0:13
  heard of them, where have you been in the past year?
• 0:16
  And obviously AI native biotech, really interesting model.
• 0:22
  And the people story behind the team is incredibly interesting
• 0:26
  and exciting. So yeah, really great to get
• 0:29
  Emily on to discuss that. And welcome to the pod.
• 0:33
  Thank you so much, glad to be here.
• 0:35
  Awesome. Nice.
• 0:37
  I guess just to start off, Emily, it's always helpful to
• 0:39
  get a little bit of a overview as to your journey and talent,
• 0:43
  but also biotech as well. You know, my, my journey into
• 0:47
  talent, I actually started off as a chemical engineer years and
• 0:50
  years ago and that sort of run through the entirety of my
• 0:54
  career. I kind of look at, you know,
• 0:56
  it's a different box that you're drawing around a problem and so
• 0:58
  there's inputs and outputs and kind of process optimization.
• 1:02
  1st 10 years I recruited on the agency side, primarily
• 1:05
  statistical modellers for financial services companies, so
• 1:08
  very specific and then transitioned over about I think
• 1:13
  11 years ago now and for the first few years was in startups.
• 1:16
  So where will start up cybersecurity startup, but I've
• 1:19
  been in biotech for the last nine years.
• 1:21
  So was that Bluebird Bio during a pretty large scaling.
• 1:24
  We started with 300 people. We grew to about 1000 in 2 1/2
• 1:28
  years, spent some time at Flagship pioneering kind of
• 1:31
  building up the system to hire across 30 companies at the same
• 1:34
  time Generation Bio and other gene therapy company and then
• 1:39
  are a capital. Prior to formation I LED talent
• 1:43
  for our incubator, but I've been at formation now for about 6
• 1:47
  months and really happy to be here.
• 1:49
  Nice, what my formation like when I feel a little bit about
• 1:52
  the thinking I'm joining them. How did it come about?
• 1:56
  Well, it's interesting because when I left, Rai had started a
• 2:00
  consulting for company. I was going to help multiple
• 2:02
  small companies. And during that time with that
• 2:05
  space, I got really into the capabilities of AI, just having
• 2:08
  a little bit of space to kind of catch up, not that we're ever
• 2:10
  caught up to where things are. And so with formation, I feel
• 2:14
  like it just combined all of the things that I'm interested in,
• 2:17
  you know, a scaling company where culture is extremely
• 2:21
  important and getting that right is going to be critical.
• 2:24
  But also the fact that it's AI, meaning I get to have access to
• 2:29
  tools and, and get to kind of at scale, continue to work on the
• 2:33
  things that were interesting me during that short time.
• 2:37
  It's an amazing team, not just the broader team, but my talent
• 2:39
  team is fantastic. So it was one of those those
• 2:42
  sort of opportunities that you kind of have to just jump on.
• 2:47
  Yeah, couldn't turn it down, that's for sure.
• 2:49
  Like you said, you've been with the business six months.
• 2:51
  I'm sure what's already happened and and changed in that time for
• 2:55
  those who've been living under a rock, like I say at the start,
• 3:00
  run us through a little bit about the model and and what you
• 3:03
  guys are doing just as a high level.
• 3:06
  Yeah. So you know our our model is
• 3:08
  evolved over time. We were initially Trial Spark.
• 3:11
  Our goal is to be, you know, the first AI native big pharma
• 3:15
  company, but we're going to be crawling, walking, running.
• 3:18
  So for a period of time, we were building technology to support
• 3:21
  clinical trial acceleration. You know, our entire goal is to
• 3:24
  make drug development faster and more efficient.
• 3:27
  A few years ago, we rebranded from Trials Park to Formation
• 3:31
  Bio and started in licensing assets of our own and developing
• 3:35
  them. So really, I'd say at this
• 3:37
  point, we have 3 built pillars to our business.
• 3:40
  So there's the drug hunting, you know, selecting the assets,
• 3:44
  figuring out what we should in license and develop.
• 3:46
  Then there's our R&D strategy, sort of how we're going to
• 3:49
  approach development. And then there's clinical trial
• 3:52
  execution and design. We got a lot of wraps in terms
• 3:54
  of running clinical trials while we were trial Spark.
• 3:56
  And so leveraging that in that in that way as well.
• 4:00
  Nice. And AI is obviously built in
• 4:03
  across all of those pillars, right?
• 4:06
  Yeah. Yep, Yep.
• 4:09
  Everything from bringing the entirety of the clinical
• 4:12
  trial.gov data set into one of our models to help our our
• 4:16
  business development team understand not only why an asset
• 4:19
  might be interesting, but also what indications we might apply
• 4:21
  it in and how that clinical trial will be run to obviously
• 4:26
  clinical trial efficiency and optimization.
• 4:28
  That's something that we've been doing for a long time.
• 4:31
  But really I think the drug development aspect is, is kind
• 4:34
  of the newer space where we're trying to leverage AI to help
• 4:38
  us, you know, think about drug development and, and ways that
• 4:40
  we can approach all of the components of translational
• 4:44
  medicine, you know, clinical development where we can
• 4:47
  leverage AI to make us smarter and more efficient.
• 4:50
  And that that's the key, right? Higher efficiency, quicker
• 4:54
  getting to the next value inflection point for any of the
• 4:57
  programs that you're involved in, whether it's preclinical,
• 5:00
  phase one, get into that next point quicker that it will be
• 5:04
  historically. It's a good point.
• 5:06
  So most of the assets of being licensed are already already
• 5:09
  have some critical data. So phase two, typically,
• 5:12
  sometimes phase one may go earlier at some point.
• 5:15
  But yeah, thinking about getting to the value inflection point.
• 5:18
  But more than that, trying to think about, you know, if you
• 5:20
  look at the bottleneck of drug development, drug discovery has
• 5:24
  been, you know, improved a ton by AI.
• 5:27
  It's really cool. But the bottleneck is really now
• 5:29
  like developing those drugs. There's more drugs that have
• 5:32
  been discovered that can be developed.
• 5:33
  So the more efficient we can make a process that's only
• 5:36
  become longer and more expensive over the last 25 years,
• 5:40
  obviously the faster we can get to inflection points.
• 5:42
  But also hopefully that savings can be passed belong to patients
• 5:47
  and lowering drug prices overall.
• 5:49
  Yeah, absolutely. I mean, yeah, that that has been
• 5:52
  the bottleneck for for years. I guess there is a somewhat of a
• 5:55
  limit to how fast you can speed up clinical trials given that
• 5:58
  they need time to to take place. But yeah, it's it's a really
• 6:02
  interesting approach. I guess going back to your side
• 6:07
  on the talent acquisition strategy with it being such an
• 6:12
  AI native company, how are you sort of testing for that when
• 6:17
  you're looking at your powering plans for the for the year
• 6:21
  ahead, you're looking at some key positions.
• 6:24
  It obviously it's not your classic big pharma or even
• 6:28
  biotech type environment, right? It's, it's a lot different.
• 6:33
  How are you going about that with the talent team?
• 6:35
  I mean it's surely a multi pronged approach I'm sure.
• 6:39
  No, absolutely. And it's, that's one of I think
• 6:41
  the more interesting things, you know, because I part of
• 6:44
  organizations that have had like a machine learning or AI
• 6:46
  component to them and also had, you know, the, the drug
• 6:49
  discovery side of things. And now we're sort of
• 6:52
  approaching it for drug development.
• 6:53
  And I think really what we're trying to be aligned on is the
• 6:57
  people that we bring into the organization, regardless of the
• 6:59
  role, really need to be aligned with our values.
• 7:02
  And, you know, our values kind of inform how we think, how we
• 7:05
  approach problems, how we engage in back and forth discussion and
• 7:09
  debate. And so making sure that people
• 7:12
  have that capability have, you know, people who are are sort of
• 7:15
  wired that way find it invigorating.
• 7:16
  People who aren't would probably find it pretty miserable.
• 7:20
  So what we've been looking at from a talent acquisition
• 7:22
  standpoint is, you know, how are we asking questions?
• 7:25
  How are we evaluating the data that we're getting back from
• 7:28
  candidates? I spend a lot of time looking at
• 7:31
  transcripts from our interviews and really trying to understand,
• 7:34
  you know, what is it that predicts whether someone is
• 7:37
  going to have a successful outcome?
• 7:38
  We're not making any decisions, but it's helpful to have, you
• 7:41
  know, one of the things I think AI is amazing at is summarizing
• 7:44
  large data sets. And we record all of our
• 7:46
  interviews. So you have transcripts, you
• 7:48
  have scorecards. So really trying to understand
• 7:51
  and we modeled our new talent strategy, our 2026 talent
• 7:54
  strategy around this is, you know, sussing out those value
• 7:57
  signals at every stage of the interview process.
• 8:00
  So we do still ask company competency based behavioral
• 8:03
  questions like design them such that they can also elicit a
• 8:07
  specific value signal. And, and that's right now, I
• 8:11
  think being rolled out, I, I feel like we're seeing some
• 8:13
  success early, but it's one of those time will tell.
• 8:16
  And AI, you know, just continues to get better and better.
• 8:19
  Our CTO always says today is the worst that AI is ever going to
• 8:22
  be because it's any aims to improve.
• 8:25
  Yeah, true. A bit of a scary thought as well
• 8:28
  because it's already pretty good.
• 8:30
  It's interesting you mentioned the value based sort of approach
• 8:35
  to to high rate and you touched on that.
• 8:38
  It's been a lot of manual work as well going through those
• 8:41
  transcripts. How have you found the AI tools
• 8:44
  that you use? How, how well are they picking
• 8:47
  out the, the answers? Because obviously a lot of it's
• 8:51
  open to interpretation, right? A conversation you're having,
• 8:54
  not just a buzzword, how, how people respond.
• 8:57
  It's yeah, it's a bit convoluted with that.
• 9:00
  No, and that's been one of the more incredible things actually
• 9:02
  about being here. So you know, when I first
• 9:06
  joined, I was, you know, like you said, manually pulling in
• 9:09
  transcripts, looking into chat, one of the ChatGPT asking it to
• 9:12
  give me themes and and things that maybe we could probe deep
• 9:15
  more deeply on. And now we actually have a tool
• 9:17
  that our internal team has built called ARC, where it's an MCP
• 9:21
  hub and I'm able to talk directly to the transcripts and
• 9:24
  directly to the scorecards. And so I no longer there have to
• 9:28
  be copying and pasting and things like that.
• 9:30
  I also am able to build artifacts where I share with it,
• 9:33
  you know, our company philosophy, our values, how we
• 9:36
  evaluate values. And so I can give the AI model
• 9:39
  the context around what it is I'm looking for the AI to
• 9:43
  understand about those transfers, about those four
• 9:45
  cards, you know, all the way down to like a thematic analysis
• 9:49
  of the people that we've hired over the last year that have
• 9:51
  been successful and seeing what kind of traits do they have,
• 9:54
  what did we learn about them in their interviews.
• 9:56
  And a lot of it has become a lot less manual.
• 9:59
  So I can do quite a bit more and my team does this as well.
• 10:03
  So we can just get better partners to the business.
• 10:05
  You know, I, I think, I think we're lucky in that our, our
• 10:08
  recruiters are much more than order takers.
• 10:10
  They're very much strategic partners embedded in the
• 10:13
  business understanding, you know, not only the role itself,
• 10:16
  but the impact the broader organization and, and how people
• 10:18
  sit. So yeah.
• 10:21
  Yeah, sure. I guess it's sort of a snowball
• 10:25
  effect as well, right? Because the more data you can
• 10:27
  give those AI tools, the better they become a picking amount.
• 10:32
  So yeah, you've done six months, but in six months time it'll
• 10:36
  just be, yeah, exponentially better, in theory at least.
• 10:40
  Yeah, and I think the next stage in where we're trying to work
• 10:42
  now is more agentic. You know, instead of me saying,
• 10:45
  hey, I'm curious how this interview went, I'm going to go
• 10:47
  ask my, my P hub. I would love that.
• 10:51
  You know, an agent could then go and just be doing that for me
• 10:54
  and saying, hey, here, here's a candidate that just went
• 10:56
  through. Here's our evaluation.
• 10:58
  Here's some questions to maybe ask in the next round.
• 11:01
  And and so my my next step is, is more automation around this.
• 11:06
  I think agentic AI holds a lot of promise.
• 11:09
  And is that the team internally they'll still be building it or
• 11:13
  you utilizing? It Yeah, yeah.
• 11:15
  I mean, again, we're so lucky to have so many resources.
• 11:18
  I have been leading talent acquisition in a number of
• 11:21
  organizations over the last 10 years.
• 11:23
  And often I found myself like really trying to make the case
• 11:26
  for a tool or a system or, you know, a solution.
• 11:29
  And here it's like the opposite. You know, we did a whiteboard
• 11:32
  session, for example, last week with our product team there.
• 11:35
  Our recruiters sort of did a pie in the sky analysis of our
• 11:37
  entire process and like, where are the friction points and
• 11:39
  where could we use help? And they're now off figuring out
• 11:42
  things to propose to us that could help us.
• 11:44
  It's it's, it's fantastic. And and that happens in our
• 11:47
  group, but it also happens, you know, across every functional
• 11:50
  area. Yeah, nice.
• 11:52
  I know something that we spoke about before is that and you
• 11:56
  were came to get across it's it's not there to replace the
• 11:59
  human element that the decision making process for you guys.
• 12:03
  It's, it's just there as a supporting function, right?
• 12:06
  It's it'll never replace that for you guys, I'm sure.
• 12:10
  No, no, no. And the goal really is to allow
• 12:13
  the recruiting team to be developing those personal
• 12:16
  relationships more clearly. You know, again, I think that
• 12:20
  that AI, where it stands right now in terms of its function in
• 12:23
  the talent space, at least where we're leveraging it the most, is
• 12:27
  to really get insight across a lot of different data points,
• 12:31
  help us really get a full, complete picture of a candidate
• 12:34
  and kind of where there's alignment and where there may
• 12:36
  not be alignment. So, you know, we're not just
• 12:38
  going by our gut. We're not just saying, oh, I
• 12:40
  like this person or, you know, leaning in on bias.
• 12:42
  We're very structured about the interview.
• 12:44
  So yeah, the goal is to lead to more time to develop
• 12:47
  relationships, really. Yeah.
• 12:49
  And I guess that's the key for the home managers throughout the
• 12:51
  process as well, right, Because they can spend more time asking
• 12:54
  targeted questions. Exactly.
• 12:56
  Feedback. I mean, yeah, you and I both
• 12:59
  know having been in agency recruitment, like the amount of
• 13:02
  fairly shortly interview processes out there, yeah,
• 13:06
  they're pretty high. So anything that can be done to
• 13:08
  improve that for candidates as much as anything, that's great.
• 13:11
  Well, I mean, and every, there's only a limited amount of time,
• 13:13
  right? Like so most interviews are 3045
• 13:15
  minutes long, an hour at the most.
• 13:17
  So making you're spending that time in the most useful way for
• 13:20
  you and for the candidate. I think is is really critical
• 13:24
  and AI can just help us ask better questions.
• 13:27
  Yeah, for sure. I think we've all been there
• 13:29
  where a candidate's met four or five different leaders in the
• 13:32
  business and has had the same conversation four or five times
• 13:36
  so. My hope is that that doesn't
• 13:38
  happen here. I think we have a, like I said,
• 13:40
  we have a pretty structured process at the beginning of our
• 13:42
  hiring process. We map out what do we want to
• 13:44
  learn, what are critical key competencies and then we assign
• 13:47
  competencies to each interviewer.
• 13:48
  And you know, everybody has a role to play.
• 13:50
  And I, I love that because I think, I hope and please, if
• 13:54
  anybody out there has feedback otherwise that candidates are
• 13:56
  having having a unique conversation with every person
• 13:58
  here. Yeah, it definitely helps to
• 14:01
  keep them more engaged throughout the process as well
• 14:03
  and paints a much better picture on how organized and how on it
• 14:08
  you guys are in turn. It's the first window into a
• 14:10
  company, right? Like kind of shows like how we
• 14:12
  operate and it should reflect what it's like to be here.
• 14:15
  Yeah, absolutely cool. I know we've touched a lot on
• 14:19
  the the positives. I guess what are some of the
• 14:21
  limitations that you've you've found so far?
• 14:24
  I think when we spoke initially, it was maybe on the sourcing
• 14:27
  front, but that's somewhat changing as well.
• 14:30
  How? How fast it's moving?
• 14:33
  Well, I mean, I, I, I do still think that there are areas where
• 14:37
  AI capabilities are not there yet.
• 14:38
  Yes, I think for sourcing, and I don't know how much of this is,
• 14:42
  is it's just a bias for us to want to look at every resume and
• 14:46
  want to just be extra sure that we're, you know, I, I love that
• 14:49
  about our team. Like we just want to make sure
• 14:52
  that a human is looking at all of the resumes.
• 14:55
  And so there's always going to be a hesitation around, you
• 14:57
  know, letting AI do that for you until there's a lot of
• 15:01
  confidence. I think also, you know,
• 15:03
  scheduling seems like such a ripe area for AI to be able to
• 15:06
  help with. I haven't seen it be effective
• 15:07
  or successful in a way that we can just let it run.
• 15:11
  You know, automation tools have helped, but I think we've got a
• 15:13
  long way to go, maybe shorter than I think, given how fast
• 15:16
  everything is changing. I think to your point earlier,
• 15:19
  like it's really important that how you're prompting the AI is
• 15:22
  not to say, hey, should we say yes or no, but to say, hey, give
• 15:25
  me insights on what more I might want to learn and some theme
• 15:28
  thematic presentation because, you know, everybody's different.
• 15:31
  Every rule is not black and white.
• 15:33
  So, you know, just making sure that we're not over relying on
• 15:36
  AI and I don't think we do, but I, I, I think that that's
• 15:40
  something that can feel like a good idea and I don't think
• 15:43
  it's, I don't think we're anywhere near that.
• 15:46
  Yeah, it's it's interesting that you mentioned about wanting the
• 15:49
  human pair of eyes on it. I can certainly empathize
• 15:52
  without myself. Like I think there's AI is
• 15:56
  absolutely universal now, right? Everyone has access within Yeah
• 16:00
  for any reason the amount of or the possibility for someone to
• 16:06
  you know, completely get a full profile full LinkedIn, full CV
• 16:11
  conducted by ChatGPT or cloud or whoever it's almost AI review.
• 16:19
  This is AICV right? You need that human touch to it
• 16:23
  as well, which I think is why what you guys are doing with the
• 16:27
  decision making is yeah, it's absolutely paramount for sure
• 16:31
  nice. I know you touched on when we
• 16:33
  spoke initially that sourcing is somewhat changing and that you
• 16:37
  maybe looking at agents for, you know, you know, what ex company
• 16:41
  did this study or went through this process at this time?
• 16:45
  Can you go out and find folks who sort of were involved in
• 16:48
  that department that time? Is that something that you'd
• 16:52
  foresee coming up in the near term or?
• 16:55
  Yeah, No, I think so. And we've done this so far, I
• 16:57
  think in a way that leverages AI, where AI is good so far.
• 17:02
  You know, for example, if we're looking for companies that are
• 17:06
  roughly our size who have enlightened assets successfully,
• 17:10
  a deep research can do that really, really well across a
• 17:13
  number of the platforms. So that can give us a starting
• 17:16
  point of where to go. But I think that there's still a
• 17:18
  lot of, you know, meticulous combing through to see who those
• 17:20
  people are and, and to know who to reach out to and to have a
• 17:25
  compelling, you know, message that's specific to them.
• 17:27
  But it certainly gets us a long way closer to what we're trying
• 17:31
  to get to. So, so we are leveraging it.
• 17:35
  I mean, again, I have like these dreams in my mind of like, you
• 17:38
  know, as we're, as news is coming in and, and there's more
• 17:42
  information like having that feed into our, our talent engine
• 17:45
  and having us have a better understanding of like who might
• 17:48
  be open to a new opportunity, what kinds of experiences people
• 17:50
  may have recently gained. You know, years ago it was like
• 17:54
  static resumes in your applicant tracking system.
• 17:56
  Now enhancement is pretty, pretty standard.
• 17:59
  Being able to layer on top of that, like industry trends and
• 18:03
  actual day-to-day industry things, I think would give us a
• 18:06
  leg up, but I don't think we're there yet.
• 18:09
  Yeah. I mean, it's interesting that
• 18:10
  you mentioned that the guys have built a tool to work with
• 18:13
  clinical trials that Gov I mean, yeah, that could be something
• 18:16
  for for us as well, I'd say. Nice, Awesome.
• 18:20
  And then just back to I guess the bigger picture on you guys,
• 18:23
  what can we sort of expect in the next, Yeah, three to six
• 18:27
  months in terms of, you know, new milestones, Yeah, new assets
• 18:32
  coming in. I'm presume you have to keep
• 18:34
  fairly tight lipped or not, but. But I think we're, we're
• 18:38
  relatively transparent. You know, we have a goal of, in
• 18:41
  licensing 3 to 4 assets a year and hat, you know, once we get
• 18:44
  to 10, we'll give us ten shots on goal.
• 18:46
  I I think that we're in a position where if one of those
• 18:49
  shots is successful, then then we've done well.
• 18:51
  If we meet the industry average of three out of 10, then we've
• 18:54
  done really, really well. And if we're able to overfit it,
• 18:56
  then we're in a great spot. So I think right now we're
• 18:59
  really focused on developing the the assets we've been licensed
• 19:02
  recently. So they're three in the last six
• 19:05
  months that were brought in by our BD team and well on track to
• 19:08
  you know, more this year. And you know, our team is just
• 19:12
  trying to make sure that we're ready.
• 19:14
  You know, we have in terms of how we figure out what our
• 19:16
  hiring is going to look like, it's obviously tied to success
• 19:19
  and in licensing these assets. So we're very proactive about
• 19:22
  seeing like I have a window into what's happening on the BD team
• 19:25
  and understanding getting close and then what roles would need
• 19:28
  to be unlocked and backing down ahead of that.
• 19:30
  So, you know, I think it's, I think it's kind of a flywheel
• 19:34
  that we're building. But yeah, I think in the next
• 19:37
  the next year you'll see, you'll see more from formation for
• 19:39
  sure. So what's that at least a couple
• 19:42
  more this year to expect from you guys for sure in terms of
• 19:45
  deals announced? Cool.
• 19:46
  I think so. Yeah, told me to.
• 19:48
  But that is they're certainly working hard and I think that
• 19:52
  they have a really good little, really good leg up on how to
• 19:56
  evaluate these assets given the AI component.
• 19:59
  Yeah, for sure. Nice.
• 20:00
  Well, look, it's certainly an evolving field.
• 20:03
  I'm sure it'd be much different if we have this same
• 20:06
  conversation in, yeah, six months time.
• 20:09
  So maybe we'll have to rekindle this then.
• 20:11
  But it's been a pleasure having you on, Emily, and you take your
• 20:14
  time. Yeah.
• 20:16
  Thank you so much.