Good Morning, HR #266 How HR is Leading AI Transformation at a Major Tech Firm — Transcript
Full transcript
- 0:00One of the most important things that we
- 0:01all have to answer to as leaders when it
- 0:03comes to AI isn't just what can it be
- 0:07used for or how can we get better
- 0:10business outcomes through AI. We have to
- 0:12presume that as we do that we're going
- 0:14to get a bunch of uh kind of human time
- 0:16and effort back. And one of the most
- 0:18important questions we need to be
- 0:19thinking about as leadership teams is
- 0:21what are we going to do with that?
- 0:27>> Good morning HR. I'm Mike Coffee,
- 0:29president of Imperative, Bulletproof
- 0:32Background Checks with Fast and Friendly
- 0:34Service, and this is the podcast where I
- 0:36talk to business leaders about bringing
- 0:38people together to create value for all
- 0:40stakeholders. And I'm joined today by
- 0:42Brandon Samu. Brandon is the chief
- 0:44people and AI transformation officer at
- 0:47Zapier, an AI orchestration platform
- 0:50powered by a fully remote team across 40
- 0:52plus countries. Brandon specializes in
- 0:55building high performance teams through
- 0:57talent and technology. He brings a
- 0:59diverse background spanning business
- 1:00development, finance, talent
- 1:03acquisition, talent development,
- 1:05education, and venture capital. Welcome
- 1:07to Good Morning HR, Brandon.
- 1:09>> Great to be here, Mike.
- 1:11>> So, let's just start with what Zapier
- 1:14and until we use that Zapier a lot in
- 1:18our organization. Uh it's a great tool
- 1:21and until we had this conversation I've
- 1:23always called it Zapier and so uh and I
- 1:26think probably a lot of people do so it
- 1:28makes more sense because we do call them
- 1:29zaps when we're doing the you know the
- 1:31calls. Uh so but tell us for those who
- 1:34aren't as nerdy and you get have their
- 1:37fingers as dirty as mine in the
- 1:39technology side of their business what
- 1:41does Zapier do? Well, you know, whether
- 1:45we work in a big organization or a small
- 1:48organization, we all use a lot of
- 1:49different software tools to do the
- 1:51things that we do best. And the reason
- 1:53why so many folks use Zapier, and by so
- 1:56many folks, I mean several million uh
- 1:58every month is to connect all of the
- 2:00applications that our teams already use
- 2:03and then automate work across all of
- 2:05them. It's used in marketing and sales.
- 2:07It's used in product development. It's
- 2:10used in uh things related to HR and
- 2:11talent. We know what I wake up thinking
- 2:13about every day in finance and
- 2:14accounting and the list goes on and on.
- 2:16But yeah, the best way to think about it
- 2:18is as a way to orchestrate work across
- 2:20all the tools you already use in a way
- 2:22that's safe uh and that also takes
- 2:24advantage of all the different ways your
- 2:26team may be using AI as well.
- 2:28>> And so and and at Zapier your title is
- 2:32chief people and AI transformation
- 2:36officer. Um so talk about that
- 2:38combination. what is what what's the you
- 2:41know why is the AI transformation
- 2:44part of your title and and what does
- 2:46that really mean in practicality
- 2:48>> you know it's interesting Mike if you
- 2:50think about the the head of people or
- 2:52the chief people officer role you know
- 2:55I've been doing this work for almost 10
- 2:56years now as a CPO across multiple
- 2:59companies and you know sometimes the way
- 3:00I describe the the ultimate mission of
- 3:02being a CPO is designing how work works
- 3:05in an organization
- 3:07the thing about that is in a lot of the
- 3:10more traditional flavors of a chief
- 3:12people officer role. The CPO doesn't
- 3:14necessarily have authority over the
- 3:17technology piece of how work works.
- 3:19These days, especially with AI, that
- 3:22technology piece is a lion share of what
- 3:24you know leadership teams are waking up
- 3:26thinking about as it relates to how the
- 3:28work works. And so at Zapier it was
- 3:30intuitive for us to have the person and
- 3:32the team who was already thinking about
- 3:34the design of work and everything
- 3:36related to talent and human development
- 3:38also thinking about how to use
- 3:40technology especially AI um given the
- 3:43opportunity.
- 3:45>> So prior to AI do you think HR should
- 3:50have had a bigger role? I mean, you
- 3:51know, we've always had an HRIS function,
- 3:53but often
- 3:55they're uh even the HRS function is
- 3:58really more IT than HR and and often
- 4:02HRIS defined processes for us rather
- 4:06than uh optimizing how we do work. We
- 4:09were kind of constrained by the
- 4:10software. But so do you think this was
- 4:13always something that should have been
- 4:14there and and just AI for some reason
- 4:16has has made that combination or that
- 4:20necess necessity for that combination
- 4:21more clear?
- 4:22>> I think that's always been on the table.
- 4:24Uh and you know despite being a chief
- 4:28people officer I I don't always look to
- 4:30uh reporting lines as a way to solve a
- 4:32problem. And so I think the best I could
- 4:34say with confidence is that way before
- 4:36AI, it probably would have been
- 4:38effective for HR teams to have a tighter
- 4:41partnership with their IT teams or their
- 4:43CTOs than is often the case given that
- 4:46those two muscles together have a lot to
- 4:48do with how work actually happens in the
- 4:50company. [snorts]
- 4:52So what do you think it's unique about
- 4:53HR
- 4:55that makes them uniquely gifted or
- 4:58suited to do the AI to you know drive
- 5:03the AI transformation piece? the the you
- 5:06know the notion of AI transformation I
- 5:08think the most important word for
- 5:10answering to that Mike isn't AI it's
- 5:12actually the transformation part because
- 5:13at the end of the day uh AI
- 5:15transformation is really the work of
- 5:17figuring out how AI you know what roles
- 5:19AI should play in an organization as
- 5:21part of the overall org transformation
- 5:24so AI transformation is simply one form
- 5:27of org transformation and if you think
- 5:28about organizational transformation this
- 5:30isn't the first one organ you know our
- 5:32teams are navigating it certainly won't
- 5:34be the past and one of the things I
- 5:36really encourage leadership teams with
- 5:38when I'm working with them on behalf of
- 5:39Zapier is a reminder that as new as the
- 5:42AI piece might be none of us as as
- 5:45leadership team folks are starting from
- 5:47scratch on the transformation part and
- 5:49we can draw a lot of patterns uh from
- 5:54prior org transformations for some
- 5:56organizations that might have been the
- 5:57shift from on-prem to cloud 20 25 years
- 6:00ago could be the shift uh in terms of
- 6:02like where you staff your talent could
- 6:04be global, you know, staffing, location,
- 6:06strategy, could be entering new markets,
- 6:08whatever the case may be. The
- 6:10ingredients for successful org
- 6:11transformation are pretty consistent.
- 6:14You know, starts with clarity on what it
- 6:15is we're really trying to do here. And
- 6:17importantly, uh has a lot to do with
- 6:19making sure that an organization has a
- 6:20culture that's healthy enough to uh to
- 6:23transform. That's where a lot of org
- 6:25transformations fall apart. It's not
- 6:27typically, you know, we didn't choose
- 6:28exactly the right tool. It's typically
- 6:30not, you know, we chose the wrong
- 6:32governance framework. Most org
- 6:34transformations fail because of a lack
- 6:36of clarity in alignment about the goals
- 6:39of transformation and uh fragility in
- 6:41the culture where even if the if the
- 6:44goals and ideas are good on paper they
- 6:46don't materialize in practice and often
- 6:48times that happens at the leadership and
- 6:49culture level and so for all those
- 6:51reasons that's why it's intuitive to us
- 6:53that the people team should play if not
- 6:55the leading role like we do at Zapier
- 6:57certainly a leading role in guiding AI
- 6:59transformation. So I think in previous
- 7:02times we would have referred to a lot of
- 7:04that as change management
- 7:06uh some workforce planning culture
- 7:10awareness and building culture out uh
- 7:12leadership development getting people
- 7:14ready to lead in different in different
- 7:17kinds of change. Uh so it's those kind
- 7:20of things. Now when you say AI
- 7:22transformation
- 7:24that seems like a temporary title
- 7:28because I imagine in five years
- 7:32we won't we'll think about using AI like
- 7:35we think about sending our you know
- 7:38paying paying a bill via Zale. it's just
- 7:40going to be built into however we live
- 7:42our lives and and so do you think
- 7:45there'll be a time where AI
- 7:47transformation is not the term and maybe
- 7:49we still got that on our you know on our
- 7:51accountability chart for the HR leader
- 7:54but it's you know it just seems like a
- 7:56very temporary title uh because it'll
- 7:58seem like you know very outdated pretty
- 8:00soon. Yeah, Mike, it's neat you should
- 8:02say that. I've talked a lot about AI
- 8:04transformation over the last two years
- 8:06or so. I've talked with very few people
- 8:07who've named that. And the reason I love
- 8:10that question is because I agree. You
- 8:12know, it's it's interesting. Uh if you
- 8:14think about um kind of digital as a
- 8:18concept from like you know 20 to 30
- 8:20years ago, you had at that time for
- 8:21example an analog you would have like
- 8:23heads of digital transformation or head
- 8:25of cloud transformation or whatever the
- 8:27case may be. Um, I think having
- 8:31named uh leaders for these important
- 8:34moments or season in an organization,
- 8:37especially with the level of change
- 8:38involved, can be helpful. But like with
- 8:41cloud or or quote unquote digital, you
- 8:43saw those serve a purpose uh and then uh
- 8:46sunset. And I I suspect the same thing
- 8:48will be true with AI transformation.
- 8:51I've got a a close friend who is a
- 8:55change management leader for a large
- 8:58organization I won't name. Uh and AI has
- 9:02taken over a giant piece of of her work
- 9:06and um one of the things you know came
- 9:10down from the CEO that this is what
- 9:12we're going to do. We're going to
- 9:13implement this. We're going to automate.
- 9:15We're gonna you know and um we're going
- 9:18to train everybody do all these things.
- 9:21and her I think her biggest obstacle is
- 9:24middle management not wanting to embrace
- 9:27it um and maybe some front line but I
- 9:30mean if you know you've got to get past
- 9:32middle management if you even want the
- 9:33front line to be able to use something
- 9:34so what what did what is Zapier's
- 9:39internal AI transformation what did that
- 9:43look like uh you know chat GPT came out
- 9:45in what fall of 2022 and it's hard to
- 9:48believe we're coming up fine three years
- 9:51uh you know four years of that. So uh
- 9:54but and it was so clumsy back then
- 9:56compared to what we get now from from
- 9:59anthropic or open AI. Um but what when
- 10:02y'all when did y'all start this
- 10:04transformation? Uh you know what was it
- 10:07that that said, okay, this is important.
- 10:10We need to get on this bandwagon. And
- 10:11how what how far after like that
- 10:14November 2022 did that happen for y'all?
- 10:17>> Yep. November 2022 sticks out for me for
- 10:19exactly the reason that you mentioned in
- 10:21March of 2023.
- 10:23So call it five months later, four or
- 10:26five months later. That's when we
- 10:27actually, you know, when you're when
- 10:29you're calling a team into a season of
- 10:31rapid change, it can be really helpful
- 10:33to to make it a moment of contrast is
- 10:36what I would call it. And in our case,
- 10:38we called it a code red. It was Zapier's
- 10:41AI code red. It's kind of interesting. I
- 10:44think in the months before after that
- 10:45Google did a AI code red as well because
- 10:47they they woke up and realized they were
- 10:49uh you know behind on the on the AI race
- 10:52and as a you know that's it's like
- 10:54literally alarmist right code red and
- 10:57not everyone loved that at the time. I
- 10:59remember this at Zapier folks are like
- 11:00that's a little much or gosh it's like
- 11:02anxietyinducing.
- 11:04I do think it was really important to do
- 11:06it anyway uh because it paints contrast.
- 11:08If you want folks to really pay
- 11:10attention, if you want to have a chance
- 11:12to start shifting the focus or inertia
- 11:14of an entire organization, sometimes you
- 11:16have to do things that are especially
- 11:18high contrast. Um and that's exactly,
- 11:21you know, what Wade and our leadership
- 11:23team were doing with that code red
- 11:25moment in March of 23. So wait, your CEO
- 11:29uh is the one who made the the code red
- 11:32the AI code red alarm ring that bell.
- 11:36>> That's right. We did it as a leadership
- 11:37team, but you know, Wade was leading
- 11:39from the front there that started uh
- 11:41with a kind of companywide blog and then
- 11:43a series of companywide conversations,
- 11:45Q&A forums, and so on. And really the
- 11:47big idea there and I think a lesson I
- 11:50learned from that um that I'll really
- 11:51carry with me is I'm really glad as a
- 11:53leadership team we didn't wait for
- 11:55perfect clarity to name the code red
- 11:57because practically speaking you know
- 11:59now it's been three and a half years
- 12:02there there's no such thing as perfect
- 12:03clarity. You know the job of us as
- 12:05leaders is to provide clarity. It's
- 12:07actually one of our unique
- 12:08accountabilities in an organization
- 12:10right is to provide and maintain
- 12:11clarity. But the bar for that can't be
- 12:13perfection or you'll never communicate
- 12:15anything. And so what we were able to
- 12:18communicate at the time was that we
- 12:20could now see four or five months in
- 12:22with the first mainstream version of
- 12:23ChachiPT that this was an incredibly
- 12:25powerful technology that was getting
- 12:28better quickly and that was going to
- 12:29have a meaningful hopefully very
- 12:31positive impact on our business. And if
- 12:34we were going to ride that wave rather
- 12:37than get crushed by it, we had to really
- 12:40pay attention, right? What are the
- 12:42implications of this on our business and
- 12:43our customers? How should AI start
- 12:45showing up in our product as it relates
- 12:47to what we're already out there trying
- 12:48to do in the world? What are the skills
- 12:50and abilities that our people uh need to
- 12:52hone in order to meet the moment and
- 12:55really understand this new technology?
- 12:57That's what the code red was all about.
- 12:59>> Yeah. And I think that term code red is
- 13:02a gentler version of I I was with a
- 13:04client yesterday, a consulting client
- 13:06yesterday that they know what they need
- 13:08to do to get their organization to the
- 13:10next level and kind of out of some of
- 13:13the doldrums they're in, but they can't
- 13:16bring themselves to do it. And we were
- 13:17talking through that and and I said I I
- 13:20I pulled out one of my favorite Pulp
- 13:23Fiction quotes and I said, "It's time to
- 13:25get medieval on your own ass." And you
- 13:28know, and you've got to say, "What are
- 13:29the sacred cows here? They can't be."
- 13:31And so I'm assuming Zapier, I mean,
- 13:34y'all aren't a, you know, you're not IBM
- 13:36as far as age and heritage and legacy.
- 13:40Uh, you know, y'all are, uh, a younger
- 13:42company in squarely in, you know, this
- 13:46new world, you know, the digital world.
- 13:50Um,
- 13:52what kind of sacred cows or or
- 13:56challenges did you have that you had,
- 13:57you know, bridges you had to burn or or
- 13:59ships you had to sink to to get to the
- 14:02next level?
- 14:04>> Well, I'd say there were some some uh
- 14:07some things we needed to be okay
- 14:09interrupting at the just like product
- 14:12company level and some things we each
- 14:13needed to come to terms with at the
- 14:15individual level. So at the company
- 14:16level or product level the first the the
- 14:19most you know kind of structural thing
- 14:21is that the we could start seeing as as
- 14:23AI matured uh from that code red moment
- 14:26forward that the you know kind of
- 14:29heritage shape of our own product needed
- 14:32to shift and we we could not be precious
- 14:35about it. um the thing that we're trying
- 14:37to do in the world was a source of of uh
- 14:40like you know um stability right like
- 14:44for the reasons all the reasons we
- 14:45talked about earlier like Zapier's
- 14:46mission has not changed because of AI
- 14:48but what customers expect from us as it
- 14:51relates to that mission the shape of the
- 14:52product how it's delivered that's
- 14:54changed fundamentally over the last
- 14:57three and a half years and it's really
- 14:58important for us to you know let go of
- 15:01of previously held uh you know ways of
- 15:04thinking about that in order to you know
- 15:07build for the new shape. So that was at
- 15:09the company or product level at the end.
- 15:10>> But you didn't change your mission
- 15:12though, right? Your mission was still
- 15:13integrating integrating information
- 15:15across the enterprise, right?
- 15:16>> That's that's right, Mike. And I think
- 15:18that's I think you're I think you're
- 15:19getting to another really important
- 15:21principle that as I've seen it
- 15:23distinguishes the teams that are really
- 15:25starting to get results with AI and the
- 15:27ones that are kind of waiting for that
- 15:28to materialize is you know keeping the
- 15:31number one thing the number one thing
- 15:33you know all and this is another great
- 15:34example of where you know no
- 15:36organization is starting from scratch
- 15:38when we're talking about AI or AI
- 15:40transformation. Organizations have
- 15:41missions organizations have goals and
- 15:43strategies. Now, what's interesting, I'm
- 15:45sure you've seen this too in in your
- 15:46advisory work, Mike, is that um this AI
- 15:50opportunity is a org health pressure
- 15:52test.
- 15:54At any given point in time, there is no
- 15:56organization that is immune from having
- 15:57at least seasons where you lose some of
- 15:59that clarity.
- 16:00>> What is it we're trying to do here? Is
- 16:02this mission on paper still the company
- 16:04mission? What does excellence look like
- 16:06here? What's our northstar goal? And so,
- 16:09you know, organizations that are shaky
- 16:11on any of those components aren't ready
- 16:13for transformation because
- 16:15>> we get mission creep, right? Yeah. And
- 16:18makes it really hard to change. Yeah.
- 16:19>> Yeah. I mean, the the you know,
- 16:22transformations work when there's a
- 16:24pretty solid bedrock that you're
- 16:26building on and it's those kind of like
- 16:28timeless aspects of organizational
- 16:31clarity. And so, uh, you know, and
- 16:34you're never done. That's the thing.
- 16:35Even here at Zapier, right, there have
- 16:37been times over the last four years
- 16:39alone, right, where we've been like,
- 16:41okay, like there's this little piece of
- 16:43the overall structure of like what we're
- 16:45trying to do here that it was clear 90
- 16:47days ago now feels a little less clear.
- 16:49We got to tighten it up yet again. And
- 16:51then the the frequency with which we're
- 16:52finding we have to do that, well, that's
- 16:54gone up.
- 16:56>> That's interesting. And then on the
- 16:57individual contributor level, what what
- 16:59what challenges did you have there?
- 17:01Well, there it's you know it's more so
- 17:03about uh some really important things
- 17:05around my my role like what is my what
- 17:08is my job and there too you know the
- 17:10interesting thing is there the number
- 17:11one thing is still typically the number
- 17:12one thing whether that's um you know
- 17:15hitting or beating a sales quota or
- 17:17providing timely helpful answers to
- 17:20customer questions or uh recruiting
- 17:23amazing people into the company. So
- 17:26there's a microcosm of the companywide
- 17:27truism at the individual level. And that
- 17:29also can be a really useful source of
- 17:31bedrock or stability during the change.
- 17:34But a lot of the other things feel up in
- 17:36the air for individuals, which is like
- 17:37what excellence looks like in my job is
- 17:39shifting. The bar is going up in almost
- 17:41every case. I I need to learn uh new
- 17:45skills
- 17:47or sharpen kind of existing mindsets or
- 17:50capabilities around discernment for
- 17:52example um in order to meet the new bar
- 17:56of excellence for the work that I do.
- 17:58And then there's also a little bit of
- 18:00like identity, you know, in in someone's
- 18:02daily work. Sometimes, you know, if you
- 18:05really sit down with people and they're
- 18:06being really honest with you about like
- 18:08what the not just what the work is, but
- 18:09what the work feels like for them.
- 18:11Sometimes there are fairly mundane
- 18:14things that they do but that make them
- 18:16feel really helpful or really capable
- 18:19and you know AI and automation come at
- 18:23first for exactly those things things
- 18:26that from the outside if you're doing
- 18:27like a process review oh that looks
- 18:30really tedious. You're like that looks
- 18:31tedious. People can't possibly wake up
- 18:33every day enjoying that part of their
- 18:35job. And by the way, yes, there are
- 18:37plenty of people who are like, "Yeah,
- 18:38hitting that button or copying and
- 18:40pasting this from this spreadsheet into
- 18:42this document." Like, I really dislike
- 18:44doing that. Like that this is my one
- 18:45precious life and I I'd love to spend
- 18:47less time doing that. But for every
- 18:49couple people for whom that is true, if
- 18:50you really do the heart-to-heart, uh you
- 18:52will find a couple of folks where there
- 18:54are other seemingly tedious tasks that
- 18:57actually provide some source of joy or
- 18:59reprieve from high high cognitive load
- 19:01work or simple. Yeah. So, so there's
- 19:05this sense of kind of like identity
- 19:07energy management um that you know won't
- 19:10typically come up on the surface, but if
- 19:12you keep digging, it's there. And if you
- 19:14can see it, you can account for it in
- 19:15the redesign and how you help people get
- 19:17across the bridge.
- 19:18>> And let's take a quick break. Good
- 19:20Morning HR is brought to you by
- 19:21Imperative. Bulletproof background
- 19:23checks with fast and friendly service.
- 19:26If you're an HRCI or SHERM certified
- 19:28professional, this episode of Good
- 19:30Morning HR has been preapproved for
- 19:32threequarters of a reertification
- 19:34credit. To obtain the reertification
- 19:36information, visit goodmorninghr.com
- 19:39and click on research credits. Then
- 19:41select episode 266 and enter the keyword
- 19:45Zapier. That's Zaper.
- 19:49Quick reminder that we're releasing a
- 19:51one-page discussion starter you can use
- 19:53with your HR or leadership team to
- 19:55consider the relevance of the
- 19:56information in this episode to your
- 19:59organization. The link is in the show
- 20:01notes. And now back to my conversation
- 20:03with Brandon Simoot. And y'all are
- 20:06obviously a tech company. I mean, you're
- 20:08you're you know, you're for a lot of
- 20:10your clients, I imagine you're at the
- 20:12center of their technology. you know,
- 20:13they've got their CRM, they've got their
- 20:16ERP, they've got all these different
- 20:17programs, and you're the one connecting
- 20:19them. So, y'all are technology to your
- 20:21core, I would think, and sales
- 20:24obviously, but even that's probably
- 20:26technologyheavy.
- 20:29As you've gone through this
- 20:30transformation,
- 20:32have you found that there are roles that
- 20:36no longer exist or that change so
- 20:39dramatically
- 20:40that you don't need as many people to do
- 20:43them?
- 20:45>> I can give you one example uh of a of a
- 20:48shift. I one of the things you're you're
- 20:51getting at here that I think is really
- 20:53important for leadership teams. You
- 20:54know, one of the most important things
- 20:55that we all have to answer to as leaders
- 20:57when it comes to AI isn't just what can
- 21:01it be used for or how can we get better
- 21:04business outcomes through AI. We have to
- 21:06presume that as we do that, we're going
- 21:08to get a bunch of uh kind of human time
- 21:10and effort back. And one of the most
- 21:12important questions we need to be
- 21:14thinking about as leadership teams is
- 21:15what are we going to do with that? What
- 21:18are we going to do with that? So, I can
- 21:19give you an example of of how Zapier
- 21:21answered that question. And we can focus
- 21:23on customer support. Customer support is
- 21:25a great place to start because it's a
- 21:26great it's a place where a lot of
- 21:28organizations start using more
- 21:29automation and AI. You know have very
- 21:32high volume tasks that are also very um
- 21:35valuable. It's like really important to
- 21:37like nail a customer question like help
- 21:39them fast and and really really solve
- 21:41the issue. Um and for a lot of
- 21:44organizations the way you provide that
- 21:45support is at least somewhat
- 21:47playbookked. you know, there's like
- 21:48some, you know, repeatable themes of
- 21:50questions you get and kind of like tried
- 21:52and true ways of answering those
- 21:53questions. So, for all those reasons,
- 21:54you have a great candidate for for AI
- 21:56and automation to help. Uh, same is true
- 21:59at Zapier. You have a great customer
- 22:01support team. Over the last, you know,
- 22:0324 months or so, they have reduced the
- 22:06number of minutes it takes to answer a
- 22:07customer question by more than 50%. So,
- 22:09twice as fast. The customer satisfaction
- 22:12for the answers to those questions has
- 22:14gone up. So, it's not just speed, but
- 22:16also quality.
- 22:17and uh the resulting jobs, people like
- 22:20those jobs more. And how do we know?
- 22:22Well, in part because retention on that
- 22:24team is extraordinary and their employee
- 22:26engagement scores have gone up 30 points
- 22:28on a scale of 100 on effectively
- 22:30everything we know how to ask. So that's
- 22:33a that is a one of our better case
- 22:36studies of a successful transformation.
- 22:39That's not the whole story because while
- 22:42the number of tickets we're getting has
- 22:44gone up during that same period, we have
- 22:45more customers and the complexity of
- 22:47those tickets has gone up because we
- 22:48have more products than we did two years
- 22:50ago. We still got so much time back uh
- 22:53that we had to make a choice around what
- 22:55to do with the time. We could have
- 22:58reduced the size of the team. Some
- 23:00companies are doing that. In our case,
- 23:02we really start with upside like what is
- 23:06something we can be doing by reinvesting
- 23:08that human time and attention that would
- 23:11be uh value add to Zapier in our
- 23:13customers. And for support, if you look
- 23:15at the support team today and what many
- 23:16of the folks do, there are two entire
- 23:19new types of work that customer support
- 23:22folks at Zapier do that are of high
- 23:24value to us and customers. The first is
- 23:27uh and this is you know driven in part
- 23:29because support folks spend a ton of
- 23:31time with customers and are often the
- 23:32first to hear about what they're trying
- 23:34to do with Zappy or the platform.
- 23:36>> They do a lot of additional lead
- 23:38qualification
- 23:41pass things on to sales.
- 23:43>> Okay. So for upgrades or things like
- 23:44that.
- 23:45>> Absolutely. Maybe you called in with the
- 23:46problem but in understanding what you're
- 23:48trying to do maybe we understand there's
- 23:49a huge use case there we can help you
- 23:51more with and that gets passed along.
- 23:53good for Zapier and good for the
- 23:55customer. Similarly, uh customer support
- 23:57folks at least at Zapier are often the
- 23:59first folks to hear about bugs in the
- 24:01product. In the old way, uh support
- 24:04folks would write a ticket describing
- 24:05the problem the best they could. They'd
- 24:07pass it over to engineering. Engineering
- 24:08would have to write the fix and QA the
- 24:10fix before it gets merged into the
- 24:12codebase. Now because of the support
- 24:15team's expertise and uh coding tools,
- 24:17coding agents like cloud code or codeex
- 24:20or cursor um we have folks on the
- 24:23customer support team that can write
- 24:24fixes to a lot of bugs and instead of
- 24:26passing a ticket where the engineering
- 24:28team has to start the fix from scratch,
- 24:30the engineering team receives a merge
- 24:33request that gets quality controlled and
- 24:35can much more quickly be merged in the
- 24:37codebase. And so that that's a I think a
- 24:40pretty complete example of both like how
- 24:42from a efficiency, quality, and employee
- 24:44experience point of view, you can shape
- 24:47a mission for change with AI, but also
- 24:50how you answer to the kind of inevit
- 24:52inevitable time and focus you you will
- 24:54get back. And maybe you'll convince me
- 24:56otherwise here, but I think in most
- 24:58organizations
- 25:00it
- 25:02for I think it's inevitable that either
- 25:07headcount goes down with implementation
- 25:11of AI at a you know at a robust level or
- 25:16it doesn't grow as the company grows. we
- 25:19we keep the same headcount and we can
- 25:21get more product out the door whatever
- 25:23it is whatever our service is
- 25:26for with the you know with the same
- 25:28headcount. Do you agree or or do you
- 25:31think there am I missing something? Is
- 25:33there really a a headcount growth
- 25:37opportunity
- 25:38uh
- 25:40and or if your organization's, you know,
- 25:42at a main line of revenue and you're
- 25:45going to stay there for a period, do you
- 25:46think there's necessarily a a reduction
- 25:49in headcount?
- 25:51>> I think on a per unit of output basis,
- 25:53the trend will be fewer headcount needed
- 25:56per unit of output. So that could be per
- 25:57support ticket, per hire, per $100,000
- 26:01of revenue sold, whatever the case may
- 26:02be. Yes. So even the customer support
- 26:05team at Zapier, this case study of
- 26:08transformation has changed the headcount
- 26:09scaling metric for the team.
- 26:11>> So for every every batch, you know, kind
- 26:13of forward-looking unit of additional
- 26:15tickets we think we're going to want to,
- 26:17you know, successfully help our
- 26:18customers with our sense of how many
- 26:19additional headcount are needed, you
- 26:21know, has gone down. And so the size of
- 26:23the team is actually very similar to
- 26:26what it was two years ago. But that's
- 26:28kind of to your point, right? The number
- 26:29of tickets, the complexity complexity of
- 26:31those tickets have grown. And so the
- 26:33team in the future, even as Zapier in
- 26:35our customer base grows, will grow at a
- 26:37more modest rate from a headcount
- 26:39perspective than was true before. And
- 26:41we're seeing that in many other areas as
- 26:43well.
- 26:44>> Tell me how it's affecting HR. I mean, I
- 26:46think so much of HR's time in large
- 26:50organizations especially is spent
- 26:53answering, like you say, playbook
- 26:55questions, stuff that's readily
- 26:57available in the employee handbook or in
- 27:00your benefits guide or um, you know, in
- 27:03a procedure someplace.
- 27:05And AI in a lot of organizations is
- 27:08taking that away and it's happier for
- 27:10the employee. the, you know, I don't
- 27:12want have to call somebody in HR, leave
- 27:13a message, wait for somebody to call me
- 27:14back. I can just jump on and talk to the
- 27:17bot and it'll give me the right
- 27:19information. Um, and so
- 27:22I think there's going to be
- 27:24a an impact on on your frontline HR
- 27:29folks. So, how is how is how has HR
- 27:32changed at Zapier with uh with all this
- 27:37technology?
- 27:38Well, it's a good question. You know, we
- 27:42on the ticketing side of the equation
- 27:44have a similar story to our support
- 27:46team. Now, we answer 85 95% of employee
- 27:49questions automatically. Uh,
- 27:53you know, using existing documentation,
- 27:55personalizing it for certain
- 27:57characteristics and what have you. And I
- 27:58think that's a pretty common story we'll
- 28:00see more and more of. Um, that shift
- 28:02doesn't happen overnight. And so, the
- 28:04folks that used to do that at Zapier,
- 28:05they're they're still here. they're in
- 28:06higher level ro roles they've been
- 28:08promoted into and now they run programs
- 28:11they don't run process uh or they don't
- 28:13answer tickets uh or what have you. So
- 28:16for example we have someone we hired uh
- 28:18better part of four years ago who used
- 28:20to spend you know a meaningful amount of
- 28:21her job answering everyday HR questions.
- 28:25uh now she is using AI to help us figure
- 28:28out how to gather better signal on
- 28:30employee experience so that we and
- 28:32leaders can make better decisions about
- 28:34how to shape healthy high performing
- 28:36teams in the organization.
- 28:39>> Do you think we'll have recruiters in
- 28:41three years? I think that's the one area
- 28:44that is or we that's so tedious and so
- 28:49many of the ATS's are horrible as far as
- 28:51making your life better if you're in
- 28:52that world. uh it's marginally better
- 28:55than having all paper, which I did at
- 28:56one time in my career. But
- 28:58>> um I I see recruiters as as something
- 29:02that
- 29:0490% of that work could be automated by a
- 29:08really smart agent. Um what do you think
- 29:11the future is for recruiting?
- 29:12>> I think we're going to see I think it's
- 29:14going to be really differentiated. I
- 29:16think I think companies are going to go
- 29:17very different ways on this. And here's
- 29:19what I mean. Um I agree with you that a
- 29:21lot of you look if you break down the
- 29:23process of recruiting and then you look
- 29:24at like where can automation or uh
- 29:27agents start doing the work you can
- 29:29start circling more and more parts of
- 29:31the process. So in terms of what is
- 29:34possible like the can we question
- 29:36there's there's a lot there to think
- 29:38about now. Now some companies are going
- 29:41to go beyond the can we question to the
- 29:43should we question. And the should we
- 29:45question is kind of going to depend on
- 29:46their outcomes in their business.
- 29:49uh companies that do a lot of like you
- 29:51know high volume hiring where you know
- 29:53the they have a pretty
- 29:56like the they're not obsessed
- 29:58necessarily with like quality of hire
- 30:00maybe you know like they're just like we
- 30:01we're okay with really high rates of
- 30:02attrition. Um that's just part of our
- 30:05business. We just kind of need to get
- 30:06people in to do some stuff.
- 30:07>> Warm bodies. Yeah.
- 30:09>> Yeah. So for for that and and or you
- 30:11just have very like extremely high
- 30:13volume hiring relative to like the level
- 30:15of investment you can manage for for
- 30:17recruiting. I think on average those
- 30:19shops will uh automate or or identify
- 30:24you know a much higher fraction of the
- 30:26overall recruiting workflow.
- 30:27>> I'll give you a counter example of that
- 30:29is actually here at Zapier. So at Zapier
- 30:32um we haven't uh you know our our kind
- 30:36of like size of our talent team you know
- 30:38per unit of output per hire hasn't
- 30:41changed much in the last 24 months even
- 30:43though we're using meaningfully more
- 30:45meaningfully more AI and automation and
- 30:47there are two reasons why the first is
- 30:49actually interesting in recruiting
- 30:50because for every every opportunity
- 30:52there is for automation AI to help
- 30:53recruiters there are like two things
- 30:55about AI that make it harder and what
- 30:57we're seeing at Zapier is that because
- 30:59candidates able to use AI there. We have
- 31:02two new two new problems or two two
- 31:04problems of much greater magnitude than
- 31:06before. One, the number of applications
- 31:08we get per job is multiples higher than
- 31:1124 months ago. And the second is um all
- 31:14of those applications are increasingly
- 31:15they look and sound alike. So the whole
- 31:17application is a much poorer source of
- 31:19differentiation or signal than it used
- 31:21to be. So it's kind of like you know
- 31:24using AI to to deal with AI, right?
- 31:26That's part of part of what we're doing
- 31:28with our investment in AI and
- 31:30recruiting. But the second thing we're
- 31:31doing is we are getting a bunch of
- 31:33recruiter time back uh in you know on um
- 31:37just like more some of the more like
- 31:38recruiting tedium. That's been very
- 31:40helpful. But we're reinvesting some of
- 31:42that uh in uh areas that increase the
- 31:45quality of our hires. So this most most
- 31:47specific example I can give you is that
- 31:48we really care about references at
- 31:51Zapier. You know, in a lot of
- 31:52organizations, references are something
- 31:54that sometimes doesn't happen or happens
- 31:56inconsistently. Where it does happen,
- 31:58doesn't usually have a lot of weight in
- 32:00the decision to make a hire. It's
- 32:02appear. These days, we're trying to get
- 32:04almost 50% of the signal we want to use
- 32:06to make a hiring decision from
- 32:08references.
- 32:10But to do that, we have to make a really
- 32:12big investment in sourcing, in getting,
- 32:16and then synthesizing and accounting for
- 32:18references, right? We built our uh
- 32:20system to do that in-house because we
- 32:21couldn't find anything on the market
- 32:23that would do it that way. And so
- 32:25building and maintaining that and then
- 32:26also like just like the people power
- 32:28required to do it well like that's an
- 32:29investment we've made with some of the
- 32:31time back from automating some more of
- 32:32the traditional recruitment process.
- 32:35So that's I'm interested in that because
- 32:37obviously one of the things I do is own
- 32:39a background investigations firm and you
- 32:42know I always say references are you
- 32:44know if it's a personal or professional
- 32:45reference besides just verifying
- 32:47employment but actually you know
- 32:49actually talking to somebody about this
- 32:50person's ability somebody who would have
- 32:51firsthand knowledge I always say that a
- 32:54reference is primarily an intelligence
- 32:57test is the person smart enough to list
- 32:59three people who will say nice things
- 33:00about them
- 33:01>> but when y'all are doing those uh And
- 33:04there's ways around that too to develop
- 33:06other references that the applicant
- 33:07didn't give you and we do those on
- 33:08executive searches typically. But um
- 33:12when y'all are going through that
- 33:14process, does that mean that your people
- 33:16are spending more time on the phone
- 33:18actually talking to real people or have
- 33:20y'all automated the collection of that
- 33:22data from uh from those sources?
- 33:25>> Both.
- 33:26>> Okay.
- 33:26>> Both. So our our reference packets uh
- 33:30include references that are written, but
- 33:32it also includes a couple that are live.
- 33:34The live ones uh sometimes are just
- 33:36based on like how the referer wants to
- 33:38or how the reference person wants to
- 33:40provide the reference. But it's also to
- 33:42combat fraud
- 33:44>> because we're we're seeing a lot more
- 33:45fraud at every stage in the process. So
- 33:47any any opportunity we have to do
- 33:49something live is, you know, can be time
- 33:51very well spent. We do use the the tool
- 33:53that we've built in house to personalize
- 33:56those uh reference questionnaires to
- 33:59kind of exactly the holes we're trying
- 34:00to fill or the things we're trying to
- 34:02verify when we would like pick up the
- 34:03phone and call a reference. We try to do
- 34:05them on Zoom where we can as an
- 34:07additional degree of verification.
- 34:09And um interestingly kind of like you
- 34:12were saying and this was a little
- 34:14provocative at the time but these days
- 34:15we do those back channel references
- 34:17there's like you know the ones the
- 34:18candidate doesn't give you but that you
- 34:20think probably have good perspective uh
- 34:23in almost every for almost every search
- 34:25now in including you know more um entry
- 34:28level or career level roles. Yeah, that
- 34:31that's you get some weird responses
- 34:33sometimes, but uh because they're not
- 34:36expecting it and that's the best part
- 34:38is, you know, that's what you want is
- 34:40somebody who hasn't been prepped and
- 34:41isn't the fishing buddy or goes to the
- 34:43church with this person or whatever it
- 34:44else it is. So, that's interesting. So,
- 34:46let's talk let's wrap up with just
- 34:48talking about what y'all have a thing
- 34:50called an AI transformation center of
- 34:52excellence. talk about that and its role
- 34:54in integrating your AI practices
- 34:58throughout the organization.
- 35:00>> The the uh the team that's responsible
- 35:03for creating the conditions where a
- 35:05customer support team like the one here
- 35:08at Zapier can do what we talked about
- 35:10earlier, the conditions uh that would
- 35:13enable a sales team to take the quota
- 35:15carrying capacity of a rep from $500,000
- 35:18to $900,000 in a year. the
- 35:20accountability for upleveling business
- 35:22results. And this is I think this is one
- 35:23of the most if if we've done anything
- 35:25well on any of this. This is one of the
- 35:27most important things we've done is to
- 35:29keep accountability for results where it
- 35:31belongs. And what I mean by that is
- 35:33accountability for those support metrics
- 35:35is still with Lauren, our support exe,
- 35:37customer operations executive.
- 35:39Accountability for quoting capacity of a
- 35:42sales rep, that's still with Naveiv, our
- 35:44sales executive, and so on. Right? The
- 35:46number one things are still the number
- 35:47one things. and accountability stays
- 35:49where they've been. But the enablement
- 35:52layer, what the center of excellence
- 35:53provides, has been very important. That
- 35:56can look like things like standardizing
- 35:58investment levels on compute to make
- 36:00sure that teams aren't left with like
- 36:02these hard caps and it's like we can't
- 36:03build what we need to build and they got
- 36:04to go talk to finance. It's like we want
- 36:06to abstract that. So that's never the
- 36:07blocker. We uh we meaning the center of
- 36:10excellence created a procurement fast
- 36:12lane about a year and a half ago uh to
- 36:15make it a lot faster to get AI tooling
- 36:17through procurement while actually
- 36:18raising the level of rigor that we put
- 36:21into reviewing those tools before we
- 36:22deploy them in the organization. So that
- 36:24was a big challenge, right? Faster and
- 36:26more rigor, really important, right? If
- 36:28you want to have a culture of uh
- 36:31innovation or experimentation with AI
- 36:34and so uh as well as the work that we do
- 36:36inside the company on AI fluency, like
- 36:38what are the human capabilities that are
- 36:40needed to work with AI in an incredibly
- 36:42thoughtful way to not just produce
- 36:44better results, but do so in a way
- 36:45that's like truly sustained and and
- 36:47derisk some of the, you know, potential
- 36:49issues you have when you use AI at
- 36:51scale. Those are some of the examples of
- 36:53what the center of excellence works on.
- 36:55And we have a brilliant director of AI
- 36:58enablement Carly Gelardi. Uh who in
- 37:01another as another example of like you
- 37:02know a lot of roles in the company have
- 37:04changed over the last you know three
- 37:06years as part of the work we've been
- 37:08talking about today. Carly I hired her
- 37:10four years ago as my chief of staff to
- 37:11the chief people officer and now she is
- 37:14the the daily driver on you know so much
- 37:16of what we've talked about today from an
- 37:18enablement point of view. And so your
- 37:20this center of excellence these are all
- 37:23HR roles that are implementing this.
- 37:26>> Importantly it's cross functional. So uh
- 37:28you know Carly for example came from the
- 37:30people team but she's not an HR you know
- 37:33she's not in an HR role as as we would
- 37:35understand it. Uh we have a member of
- 37:37the finance team as part of the center
- 37:39of excellence. we have someone from IT
- 37:40and security and the list goes on and on
- 37:43because together that's how we're able
- 37:45to you know provide these you know we
- 37:47have we do have folks from the people
- 37:48team as well especially as it relates to
- 37:50uh talent development upskilling
- 37:52reskilling so but I think importantly
- 37:54it's a cross functional effort because
- 37:55you do need those various muscles or
- 37:57ingredients to come together uh to
- 37:59create these types of conditions
- 38:02>> so how has
- 38:05your
- 38:06you know continuing integration of AI
- 38:10into the organization and that
- 38:11expectation I guess that people are
- 38:12going to use AI to uh improve you know
- 38:16what they what they you know how they
- 38:18create whatever it is that they do in
- 38:20the organization. How has that changed
- 38:22how you hire people?
- 38:23>> You know about a year and a half ago we
- 38:25were asking the same question and it
- 38:27inspired us to create a effectively a
- 38:30framework for AI fluency at the company
- 38:32and we now use it as part of our hiring
- 38:34rubric. So this is called Zapier's AI
- 38:36fluency framework and uh if you remind
- 38:40me I can include a link. We open sourced
- 38:41it so we can put it in the show notes uh
- 38:43for anyone listening in here
- 38:45>> and uh it's been incredibly helpful. So
- 38:47we we screen for AI fluency using that
- 38:49framework at the time of hire and then
- 38:51it's the same framework we use inside
- 38:52the company. We're thinking about how to
- 38:54help folks sharpen up. The the thing I
- 38:56could preview for folks that want to
- 38:58look at this later is that the AI
- 39:00fluency framework at Zapier has four
- 39:02competencies and only one of them are
- 39:05builder skills like how to use the
- 39:07tools. The other three the lion share of
- 39:10the whole framework are timeless
- 39:12timeless uh metacognitive skills. What I
- 39:15mean by that uh the first one is we call
- 39:18strategy which is really kind of like
- 39:20having enough kind of context or
- 39:21business acumen to know what's worth
- 39:23doing in the first place including with
- 39:25AI. Second one's called growth. That's
- 39:28all around persistence, resilience and
- 39:29growth mindset. Incredibly important
- 39:31especially when working with AI where
- 39:32iteration is where you start getting a
- 39:34lot of the better results like you
- 39:35rarely oneshot your way to a great
- 39:37output or great application right as as
- 39:39I'm sure you know.
- 39:41>> And the um then the third one is the
- 39:42builder skills. The fourth one we call
- 39:44accountability. And accountability is
- 39:47just like really keeping the number one
- 39:49thing the number one thing. What is
- 39:50greatness look like at the end of the
- 39:52day? Accountability for all of the work
- 39:56still rolls up to a person at the end of
- 39:58the day. So no matter no matter how much
- 40:01work AI is doing towards that outcome,
- 40:04the accountability for the quality of
- 40:05that outcome and any sharp edges or or
- 40:07you know uh errors that come along with
- 40:09that still roll up to a person. And so
- 40:11we explicitly uh chose accountability
- 40:15for that fourth competency.
- 40:17>> So in an organization like that where AI
- 40:20is so important,
- 40:22let's wrap up with does that is all this
- 40:26automation and
- 40:28you know just the removing of a lot of
- 40:30the friction in processes where it would
- 40:32take two or three people to have a
- 40:33conversation to do something and we've
- 40:35figured out ways to automate it. Has
- 40:37that lessened the requirement for what
- 40:41we traditionally call soft skills or
- 40:43power skills uh in the organization?
- 40:46>> I think it makes them more important not
- 40:48less. It makes them more important not
- 40:50less. Now now one thing is true we are
- 40:52seeing this is that the uh amount of
- 40:55like crosserson coordination that's
- 40:57required to get work done that's going
- 40:58down. Mhm.
- 40:59>> That's going down because a single
- 41:00person with good context, right, and
- 41:02good standards for work with AI can
- 41:04cover more functional surface area than
- 41:06they used to be able to. And that's
- 41:07helping us move faster. It's helping
- 41:09people feel more like a not not
- 41:11necessarily autonomy, more empowerment
- 41:13in their work. We're like, I'm doing
- 41:15this thing and I don't have to like
- 41:16throw this over the fence to somebody
- 41:18and wait two weeks and then get it back
- 41:19and then I don't know like you know what
- 41:21I It's just like it can feel very
- 41:22empowering and and people know you're at
- 41:24if you're at a place like Zapier that's
- 41:26really investing in AI fluency like you
- 41:27are learning the future version of your
- 41:29job and that too can feel uh that can
- 41:31inspire confidence not just empowerment
- 41:33in the work because you know you know
- 41:35you're you're learning what you need for
- 41:37the future. Okay. So all that aside um
- 41:40you know in terms of those kind of human
- 41:41skills or what have you um I I think we
- 41:44got to stay sharp on that because you
- 41:45know at the end of the day the things
- 41:47that make for a healthy high performing
- 41:48organization haven't changed at all with
- 41:50AI. It still comes down to clarity
- 41:52around what we're trying to do and what
- 41:54excellence looks like. You know,
- 41:55cultural conditions like psychological
- 41:57safety, culture of experimentation,
- 41:59being able to sharpen each other with
- 42:01great feedback. Those all require uh
- 42:04high degrees of of what we would call
- 42:05soft skills. And in a time of change,
- 42:07they're actually more important, not
- 42:08less.
- 42:09>> That's a great place to wrap it up. I
- 42:11thanks for joining me, Brandon. This
- 42:12flew by. It
- 42:13>> was a pleasure, Mike.
- 42:14>> And thank you for listening. If you
- 42:16enjoyed this episode, please write us a
- 42:18review on Apple Podcast, Spotify,
- 42:20YouTube, or wherever you're listening,
- 42:22and then share this episode on your
- 42:23favorite social media platform. It does
- 42:26help new listeners find us. And we'd
- 42:29love to hear your thoughts at good
- 42:30morninghr.com.
- 42:32Good morning HR is produced by Paco on
- 42:35Imperatives Marketing Team. The Good
- 42:37Morning HR theme music was created by
- 42:39Steve Pegler at HR Strategies of Texas
- 42:42and I'm Mike Coffee. As always, don't
- 42:44hesitate to reach out if I can be of
- 42:46service to you personally or
- 42:48professionally. I'll see you next week.
- 42:50And until then, be well, do good, and
- 42:54keep your chin up.
About this transcript
This page contains the full transcript of Good Morning, HR #266 How HR is Leading AI Transformation at a Major Tech Firm by ImperativeInfo, generated from the public captions YouTube serves with the video. The transcript has 8,177 words across 1,190 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.