January 24, 2025 • 51 mins
In this episode, I sit down with Emily Smith, Principal and VP of Partner Success at Collegevine, to explore how AI is reshaping higher education. Emily shares:
- How AI agents are transforming admissions and student success.
- The concept of “Time to First Value” and why it matters.
- How AI can empower staff rather than replace them.
Whether you're an administrator, faculty member, or just curious about the future of education, this conversation is packed with actionable insights.
Want more insights into the Higher Ed Space?
Check-out our Newsletter.
Please leave a review on your favorite podcast player!! It helps support the show.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:01):
I am psyched to be here with Emily.
Emily welcome.
Speaker 2 (00:04):
Thanks for having me, Mike.
I'm really, really happy to behere with you.
Speaker 1 (00:08):
Same here.
You've got such a fascinatingbackground and maybe just to
sort of square the work thatyou're currently doing and the
problems that you're solving atCollegeFind can you just walk us
through your career trajectoryand how you went from serving
higher education with startupsto then working for some
enterprise solutions and thenyour current work at Collegevine
(00:30):
how that all come to pass.
Speaker 2 (00:33):
Yeah, easy question, right to be asked about yourself
and how you got here.
It's all 2020 in the rear view,but it started for me as a
senior in college and this willmaybe like warm your career
services heart, mike, but like Idon't know why, it didn't occur
to me in the second semester ofmy senior year that like I
would have to leave college andget a job and that somehow felt
(00:55):
surprising.
And I I'm a ski bum at heartand I was like, okay, well, I'm
just gonna, you know, ride offinto the sunset and and ski my
little brains out.
And then I realized no, youactually need to make money and
contribute to earth.
And at that time I had beenworking with my career services
office, who encouraged mebecause I had been a tour guide
in the admissions office at myalma mater to talk to these
(01:16):
consultants who were coming in,who kind of wanted a generalist.
And I took the meeting and I'mso glad I did, because that set
me off working for a consultinggroup doing enrollment
consulting and services rightout of college.
And I got pulled into roomswith university presidents and
provosts and vice presidents ofenrollment at 22 years old,
being asked what I thought aboutthings, which was a really
(01:38):
powerful way to feel reallyimportant early in my career.
So I got pulled along like thelittle sister of these serious
enrollment consultants, fixingserious enrollment problems, and
after that I moved on to reallythe software product side of
things.
I was the first employee at acompany called Fire Engine Red
where we really helped to inventemail marketing for colleges
(02:00):
and universities and did a lotof software products aside from
the CRM on the enrollment sideof things.
We also did a lot of sort oflike software products aside
from the CRM on the enrollmentside of things.
We also did a lot of studentsearch and the student search
when I and by that I mean likepurchasing names and marketing
to them, doing the enrollmentmarketing via purchase names of
test takers and and other otherpeople like that, and in that
era that motion like workspretty well and it started to
(02:24):
work less and less well overtime, um, and I really enjoy
being part of the small, smallbut growing thing that companies
do.
Right, I was the first employeeof a company so I got to do all
the jobs, um, and as I movedthrough my career, working, you
know, at another company doingCRM for admissions and student
success.
Again.
I joined as a smaller companywas part of growing that company
(02:46):
to be bigger and bigger and Iseem to keep going back to the
small scrappy and being therefor the really exciting growth
and I really love that because,as slow as we observe, colleges
moving companies can move reallyfreaking fast and it's really
really fun to like test stuff,fail and iterate in that cycle
and to me there's this throughline of like really figuring out
(03:09):
how to get a pulse on whatstudents serve students really
well and that's changed a lotover time and being a part of
feeling like a fundamental fixerof those problems is just
really important to like my, mypersonal mission.
Speaker 1 (03:27):
That is so cool and I appreciate you contextualizing,
ensuring all those thosedifferent pieces and those those
three lines like how do yousort of sit at the intersection
of, like those two verydifferent things?
Right, you're serving this likevery traditional industry that
tends to sort of move slow,right, where iteration is not
necessarily sort of baked intothe culture where they're
(03:50):
thinking in timelines that aresort of decades long, but you
juxtaposition that with the typeof work that you're doing,
where it's like it's quick, it'siterative and it's like, and
it's serving all these spacesLike.
how do you hold those twoextremes and find that happy
middle ground between the two?
Speaker 2 (04:07):
Colleges are not startups.
If you've ever worked for, likeon the corporate side of things
we think about, you know bigcompanies that do things like
really.
I guess a rude reading ofworking for a really big company
is like it might not beapparent.
If you didn't show up for workfor like a week, two weeks in a
row, like your impact would notnecessarily be felt.
(04:28):
And colleges are a little bitlike that in that they are
really established businessesthat we expect to be open and
available for jobs, for theproducts that they provide, a
community, and in startupsoftware we have to be so
scrappy and everybody'scontribution, like every
individual employee in a startupcompany, their impact really,
(04:51):
really matters.
So to bring some of that mindsetinto higher education, which is
like bigger, a little slowerand a little more stayed, feels
really exciting on theinstitutional side.
I think it lands really well.
I also think there's some sortof leverage and power that you
just get as the outsider right,like you might be saying
something as a consultant thatyour cabinet had been sharing
(05:12):
with you, that you're justhearing it differently from a
consultant, and I understandthat there's power in that.
And the third thing that Iwould mention is really around
like the advantages that I havesitting in my seat, peering
through the windows like this atlike hundreds of institutions.
Folks I work with on theinstitutional side.
They know their politics reallywell.
They know the ins and outs oflike how things are politic or
(05:36):
clicked through in their owncampuses.
But I get this advantage oflike bringing in this broader
outside perspective, havingworked with hundreds and
hundreds of colleges over time.
So my observations, whilethey're maybe a little bit
broader than you would give fromthe inside, get to be
transformational just because Iget to be the outsider.
Speaker 1 (05:57):
Yeah, super cool, I love that answer.
Can you walk me through alittle bit about like college
run specifically?
Um you all have leaned into,which I think is just so awesome
?
Um focused on the sort of aispace and the ai integrations
and supporting particularly sortof enrollment.
(06:18):
Um you know how.
How have you sort of like what?
What's collegeVine's journeybeen like?
And then what's been yourunique contribution?
Because you're sort of bringing, I think, a longer narrative
and historic lens about likedeeply understanding CRMs and
seeing that sort of originalmigration to email and all these
(06:40):
sort of like bigger techchanges to serve a traditional
industry.
What's it been like to sort ofbe on that front line of AI and
what's been your own sort oflike product journey at
CollegeVine to land in where youare now?
Speaker 2 (06:53):
Yeah, so I've been at CollegeVine the last three
years.
The company I'm not a founder,the company is many years older
than that and has been through acouple of different iterations,
and the way I would describewhat Collegevine contributes to
the space now is that we're aplatform for building and
deploying AI agents atinstitutions, and that is a sort
of fifth order thing that we'vegotten to over time, because
(07:15):
there are a bunch of otherthings that we do in service of
helping students and familiesget the best guidance possible
at every stage of life, whetherthat's the transition from high
school to college or beingsuccessful as a college student
or transitioning from college toa job, and there is a limit to
what humans can do, whether it'sat the sort of guidance layer
(07:38):
at all of those stages or at theadministration layer of all of
those different phases for astudent.
And now, really, as an AIcompany, we have created kind of
autonomous AI agents that goout into the world and do work
with constituents, and that'slike things that you would if
you were a college president orprovost listening to this and
(07:59):
you're thinking.
You know the ways in which youlimit the requirements that you
put on your staff or strategiesthat you create to the humans
that can deliver them.
We don't have to do thatanymore.
We don't have to put thoselimits on anymore.
We can really dream much biggerthan we have been able to in
(08:19):
the past to think about what AIcan do as a really strong point
of leverage for automatingdifferent jobs at the
administration layer of ourinstitutions.
And that's the problem thatCollegeVine is really solving
right now in the college space,while bringing along students
and families in a way that feelslike it delivers really great
(08:42):
access to guidance andconnections across that whole
landscape.
Speaker 1 (08:46):
That's cool.
Can you walk me through like afew of your favorite like recent
use cases around, like howyou're solving those sort of
massive operational pain points?
Right, because it strikes me asa you have such a unique like
product and engineeringchallenge just given, like the
sort of the data, the datalayers that you have, the
challenges with integration thatyou have with all the different
(09:10):
third-party pieces, which Ithink you've had really smart
APIs and connections to solvefor those.
But can you walk me through afew use cases and then what the
felt experience is on the sideof the user or the admin?
That's really doing it.
Speaker 2 (09:28):
Yeah for sure.
So we've got several sort ofout-of-the-box AI agents and
probably the easiest one toconceive of how it does its job
is the AI recruiter.
This is something that we arelive with at several hundred
colleges, live with at severalhundred colleges, and the AI
recruiter essentially is anautonomous agent that delivers
(09:49):
recruitment activities for theadmissions office, whether
that's pairing one to one withstudents to like you would want
a counseling staff.
If you could afford it, youwould hire lots of admissions
counselors to pair one-on-onewith students to give them an
individual journey and treat,like you, mike as an individual
student, versus like ah, mike isa guy from Wisconsin who's,
(10:10):
like, interested in engineering,so we're going to bucket him
with all the other guys from outof state who are interested in
engineering.
It's inherently a differentexperience when you're working
one-to-one with a really smartadvisor, so that's one job that
the AI recruiter doesno-transcript On day 13,.
(10:52):
We send this email and then wehave to go write all those
things and create creative for.
It Turns out AI is really goodat that job and, in fact, better
than we are, which I have somefeelings about, having spent all
those logged, all those hoursdoing that job.
But AI is better than I am atit.
And the third job that therecruiter does right now is
processes, transcripts andapplications.
(11:14):
So that's again something thathumans have done for a long time
to grid highest math or well,we take this GPA and we
calculate it in this way andthat's a really expensive
operational problem for collegesto solve.
That the AI recruiter does.
So that's how to conceive ofone of the AI agents.
But the AI agents show upacross this whole student
success journey, being anadvisor to students, a concierge
(11:36):
, to either find stuff out likehey, mike, your housing form is
done.
Like why haven't you done it?
Or hey, mike, how are youfeeling?
Like have you made any friendsRight To really get at early
indicators of how students aredoing in their current student
journey.
And the applications continueto extend through the ways in
(11:56):
which we would interact withalumni or build networks for
career services.
But at the administration layer, we're thinking about all of
the different work that we canautomate in order to do it
better, faster and bigger.
Speaker 1 (12:18):
I absolutely love that answer of the felt sense of
what it's like to sort of be anadmin working with this tool or
a student that's really sort ofon the receiving end and doing
the different pieces.
It's so fascinating what youmentioned around that roadmap
and one institutional challengeI see like again and again is
like with turnover in the spacelike being relatively high and
(12:41):
documentation on process andthings like that, not depending
on institutions sort of beingbaked hard into the culture.
You have a variance around whatthat's going to look like from
year to year, depending on who'ssort of connected and different
pieces there.
That's just going to sort ofchange and iterate, not
(13:04):
necessarily in good ways, andyou also have this sort of lack
of data capture and processcapture and all these sort of
connected pieces there.
What are the feedback loops forthat on your end to sort of
ensure, like you know,communications are sort of
working well, like what are thethings that you're looking at
institution basis and how areyou empowering partners to sort
(13:27):
of like?
do some of the assessment piecesconnected to all that.
Speaker 2 (13:32):
Is the question like, is part of that question, like,
like, how do we know it'sworking?
Like, how do we measure it'sworking?
Is that sort of at the heart ofit.
Yeah, exactly, such a goodquestion, right, because, like,
in this industry turns out funfact, we're not like shaking the
money tree and it's not likeraining down dollar bills for us
to like try stuff that doesn'twork, right, like we're at a
(13:52):
critical point on making thebusiness of the institutions
work really well.
So things that work are reallyimportant and the ways that we
look at sort of measuringsuccess is like imagine your
highest order expectation foryour best possible staff member.
And if, mike, I hired you at mycollege to work in admissions
(14:13):
for me, I would want you everyweek, every day, if I wanted you
to, to be like handing me areport card of how you were
doing.
I would want to know how manystudents you were talking to.
I would want to know how manytimes you messed up.
I would want to know how manytypes of different questions you
were getting.
I would want to take that dataand be able to synthesize it to
the faculty and creators ofprograms at our institution to
(14:35):
be like yo, we're getting askeda lot for this type of program
that we don't offer.
What are we going to do aboutit?
Like, is that something we'regoing to change the business to
accommodate?
Or look, we can very easily seethat this is no longer a
program that is offering anycreating any real genuine
interest or excitement with ourpopulation.
Do we need to change theprogram?
(14:56):
Is it profitable for us to keepthe program?
So anything that you can thinkof, sort of your highest order
expectation of what a humanstaff member would deliver you
to provide you real leverage inleading the institution, are the
types of things that the AIagents sort of bring you and
help you measure, because it'sreally like that leverage that's
(15:17):
going to allow you to be makinggreat strategic decisions for
the institution.
Speaker 1 (15:23):
Yeah, very cool, very cool.
What are some other use casesthat you're excited about?
I mean, the AI Recruiter issomething that I'm just talking
with a lot of people People areexcited about.
The feedback has been reallygood.
It seems like it's sort ofrelatively plug and play and,
despite the sort of complexityof all the interconnections that
you need to make something likethat work, it seems like it's
(15:44):
working really effectively.
What are some other use casesthat you're pumped about?
Speaker 2 (15:48):
Yeah, so the stuff that we're creating now, like
imagine, if you so I'm going toask you to like go back in your
time machine to the time thatyou were leading career services
offices, mike, and like imagineif you were sitting with a kid,
like human, you were sittingwith a student and the student
was, you know, about to embarkon their career search.
(16:09):
And this is the point ofcollege right To create people
who can go be good citizens ofearth and contribute to our
societies and cultures.
Like that's the point, and you,as the career services officer,
are trying to make that happenin the best possible way, like
you can.
Officer, are trying to makethat happen in the best possible
way, like you can.
And imagine if I gave you atool that, like allowed you to
(16:29):
connect this student.
Let's say the student was,let's go with engineering.
Let's say the student was anengineer and was really feeling
nervous about, like, what lifelooks like as an engineer.
And what if AI could help younot only identify like sure
tabular data could do this todaybut identify who from your
alumni network worked as anengineer and was open to
(16:50):
mentoring.
Like you could get that withspreadsheets and joins in a good
enough database today?
Sure, but what if AI was thenable to go facilitate that
introduction?
Because the friction is in theintroduction making in that use
case, like, really specifically,the friction is you write down
the name of a person and say tothe student like hey, look this
person up, go build arelationship with this person.
(17:11):
That is freaking terrifying fora student.
But imagine if you had an AIagent to not only identify the
best fit alum who could go dothis job for you, also indicate
their willingness to be a mentorand also then kick off this
relationship and facilitate thisrelationship between two humans
.
That is a really high leveragemove you can make as a career
(17:35):
services person.
And there are a billion usecases like that that keep the
human fundamental relationshipsor transformational relationship
building at play, which issuper important to colleges, but
adds a techification on top ofit to make it go way faster or
better.
Or if you were a professor andyou had an incoming suite of
(17:56):
students in your senior seminarand you had your AI agent or
your AI assistant telling you oh, we've got these 20 students
you really need to watch out forso-and-so.
She's really struggling withsome other classes.
So be aware of this.
This other student is going tobe a great leader for you.
This other student hassomething going on in their home
life that you should be awareof.
This other student is headingtowards a PhD in your program.
(18:18):
That would be a really helpfulhigh-leverage thing that you
could get very quickly as aprofessor.
That would fundamentally makethe way you did your job better,
facilitated by better data andtech.
Right Like sure you could wadethrough the data that we have
today to make sense of that.
You're never going torealistically do that because
it's way too high friction.
Speaker 1 (18:39):
Totally yeah, and I love the career services example
.
I deployed PeopleGrowth on acouple of campuses and it was so
fascinating to just look at thedata sets around goes to
somebody like but you have, evenif, like you execute really
(19:08):
really well these tiny, tinysteps, you lose people along the
way.
So any way that you can sort ofintegrate that like very sort of
seamlessly, like your chancesof like the person actually sort
of doing the thing, which islike making that initial contact
.
What we figured out right away,like very quick, was like if we
got the person to send onemessage, their chances of
(19:29):
actually coming back were likeexponentially higher and sort of
using the tool where we werereally high.
But like there's so many manysteps between like the person,
like getting on the platform andthen actually finding that
first person that you walk likethat you connect with that like
if you can close that gap, likethat is a massive lever in terms
(19:50):
of like the person's careertrajectory that you don't really
sort of think of.
But it's like a bunch of ministeps consolidated.
If you can solve those painpoints, it's it's huge.
So it's really cool to see thatyou're doing this in like a lot
of different ways.
Speaker 2 (20:05):
Yeah Well, one of the things that we talk about, like
on the sort of corporate sideof things I'll sort of like I'll
bring a term in from SaaSsoftware that might be
interesting to your audiencewhich is like time to first
value, and essentially, thequicker you can bring your
consumer time to first value andthat's short, as short as
possible, the better, like thebetter life of the customer,
(20:27):
whether it's, you know, revenueor success or whatever.
We're measuring over time andin your instance of like if we
could just get them to send thatfirst email, or, on the
recruitment side, if we can havethat first interaction, be
really positive.
You're right, it creates lessfriction in the rest of that
chapter of the journey.
But that time to first valuething is really interesting
thing to measure across yourconstituents to watch that
(20:51):
friction start to dissipate.
Speaker 1 (20:53):
Yeah, that's really cool.
I've not heard it phrasedexactly like that, but I love
that.
Time to first value as a metric.
I'm just curious is theresimple ways to measure that?
Speaker 2 (21:06):
I'm sure it depends on the use case.
Yeah, it super depends on theuse case, and this is something
that we talk about both in theproduct side of tech companies
as well as the client successside of tech companies, and it's
basically whatever you thinkthe first valuable moment is,
and usually it's whatever feelslike that aha moment.
So on the admission side ofthings we might be really sort
(21:29):
of hard and fast on like well,it's the application, right,
that's the thing that allows usto really gauge interest.
Things in admissions andrecruitment and enrollment are
actually way easier for us tomeasure than other use cases
across the administration layerof an institution.
But it could be something likesentiment.
It could be something like youknow, gartner does a really
interesting survey on a set ofquestions that like allows you
(21:53):
to measure the success of youremployees and one of the
questions is like have you madea best friend at work?
And you ask adults if they'vemade a best friend and they're
like no.
I'm a 45 year old man, like Idon't need to make a best friend
, but really like those types ofthings like a student, have you
made a best friend yet?
You know, like have you joineda club yet?
Are really strong indicators ofhow a student is viewing their
(22:16):
experience at your institution.
So you kind of get to call it.
But when you call it then justlike anything else, we measure,
like the point would be then tomeasure consistency, learn,
iterate, do it again, love that,love that.
Speaker 1 (22:34):
So a question that I can just think of the sort of
audience asking because I knowthis is sort of differs wildly
sort of depending on the usecases and different pieces.
But how do you build your AIagents?
Right, like because you know Ithink the basics that you know
you hear pretty consistentlyfrom AI engineers are
essentially, you're sort ofyou're feeding it a lot of data
(22:56):
and you're doing sort ofsystematic training to sort of
get it to sort of perform theways you want giving feedback
and all those kinds of pieces.
But what's that look like foryou?
Because you have such a varietyof use cases, you're, I think,
increasingly sort of gettinginto this more sort of
customized space and havingsolutions that are sort of both
out of the box but also sort oftweakable.
(23:17):
So what's like the backend ofMagic look like for really sort
of making these things work inthe way that you want them to
work?
Speaker 2 (23:25):
Yeah, that's a good question.
And the other thing, the otherelement that I'll add to it is
colleges do a lot of feelingaround in the space of like we
know we need a thing, do webuild it or do we buy it, and
there's sort of that element,too, which is like, can we do it
ourselves?
And I think what's reallyinteresting with AI is that, you
(23:47):
know, very occasionally I'lltalk to somebody about you know,
the AI recruiter, for example,and they'll just be like, well,
why couldn't we just chat GPT,the AI that you go out and ask
something of is reacting to arequest in a transactional way,
thing of, is reacting to arequest in a transactional way?
(24:08):
In 2025, we are heading into aplace of agentic AI that is more
autonomous and proactive, whereyou give it a whole job.
Yo, ai, recruit the class.
Hey, ai, take all of thecaseload of all of our 40,000
students and make themsuccessful.
Speaker 1 (24:23):
I'm so curious about this with you all because you
guys are doing some really sortof interesting work at the
intersection of building reallycool bots that are doing amazing
things.
What are some unexpectedchallenges that you've faced in
terms of sort of deploying andopening up your range of
products and things that you'redoing, and then how have you
(24:45):
worked to overcome those?
Speaker 2 (24:47):
Yeah, I think a couple of things.
One I think most of thediscourse about AI in higher
education right now is on theteaching and learning side, and
I think that's a debate we needto solve, and solve quickly, so
that's sort of thing.
One, and by that I mean, likewe can't spend time screwing
around with, likeinstitutionally, how we feel
(25:07):
about the use of AI in theclassrooms, Because if you start
to indicate, like a less thanclear stance on this, you will
start to create ripples for yourstudents.
You will start to then havethose ripples become waves,
which will make students notfeel really confident in their
participation of your brand asan institution.
(25:28):
And I guess I care more aboutmaking a stand on this than what
the stand is, Though, of course, I think the stand is look, if
we're trying to createespecially young people who are
going to participate in theworld in meaningful ways and get
jobs, then one of the thingsthat we need to do is to train
(25:49):
them how to use AI well.
It's a job skill.
So of course, I have a verystrong opinion about how
institutions come down on this,but I actually care more that
institutions have a reallystrong stance on it.
I think, once we can let someof that air out of the room on
the focus that we spend on AIfor teaching and learning, we
need to quickly turn ourattention to what AI can solve
(26:13):
at the administration layer,Because I think about
institutionally if you correctfor hospital revenue and if you
correct for some of the sort ofenrollment, demographic, student
search cliffs, it's about a $93billion deficit on the
administration layer thatschools and what jobs we would
want to have AI automate.
(26:45):
Then I think it really frees usup to think about what
challenges and strategies thatwe might want to solve.
And one of the surprises that Ihave is the difference between
leadership at an institution.
Some of them are thinking like,okay, yes, actually, our human
costs, our cost of the humanstaff, are actually way too
great.
We need to pare that down andhave AI take over the work.
(27:07):
And then the other camp, whichis no, protect humans and jobs
at all costs.
And I sort of say to them like,okay, great, but let's think
about the ways in which using AIbetter at the administration
layer can give your existingstaff a promotion, and by that I
mean like if you use AI well,you are essentially giving
yourself a promotion.
(27:28):
So if you used to be the personwho compiled data, AI should do
that for you and you can be nowthe person that analyzes the
data right.
There's like a leveling up thatwe can offer to our staffs.
So one of the surprises I haveis sort of like how much time
we're spending on this teachingand learning debate, which I
think needs to end.
And the second is around likethis divided road on should we
(27:53):
automate people out of jobs orshould we automate enough jobs
so that people can be leveled upin their jobs, and those feel
like really fundamental debatesthat feel important to resolve
in the next chapter.
Speaker 1 (28:07):
Yeah, I think they're definitely sort of huge debates
and I completely understand thethinking through the second or
third order effects of like hey,if we're downsizing our staffs
in this way, this way, this wayor this way, you know we're
potentially looking at a verysort of different workforce and
I totally get that angle of.
(28:28):
At the same time, I meanespecially at like a private
institutions or I think youcould almost say it's sort of
non-community colleges, like thecosts to the end student is
paying are so high, right, and Ithink we are in an environment
where good, bad or indifferent,the signaling from the federal
government at this point is youknow, there's not more federal
(28:51):
dollars that are going to likean influx, that are sort of
going to be coming.
I think we've also seen thefolly of a lot of institutions
really relying on their donorbase when we know those dollars
from a outside of like a smallpercentage are going down,
giving rates are sort of goingdown.
So I think that, like I wouldargue, the writing is on the
(29:14):
wall in a lot of ways that likehey, we don't have this influx
of cash that are coming toundercut the cost of like really
sort of growing administrations.
How do we do this work moreeffectively?
Um, and I think you all solvelike a really sort of
interesting thing there.
Um, for for you like what, whatare like some of the like when
(29:37):
you're looking like two, threesteps ahead, what are like some
of the really cool things thatyou think your tech could sort
of jump in and solve that likepotentially.
I don't want to say likeeliminate jobs, but just
heighten the experience of likehow a person sort of sitting in
this sort of like adminfunctional role could maybe sort
of do the work differently.
Speaker 2 (29:58):
Well, I think about.
I think about that mindsetright, that like more money is
not coming, and I think thatbreeds innovation.
And we can choose to beflattened by that and feel
really scared about what thatmight mean, or we can choose to
let that make us curious andhungry to really like flex and
(30:19):
do different things.
And one of the things that Isee happening in the world of
work that I don't know ishappening in colleges yet is a
culture of like efficiency andfiguring it out right, like in
our jobs, and I think about theways that we, you know, like
create our work culture atCollegevine, which we've done
really, really intentionally,just in terms of how we do work.
(30:42):
There's a really new sort ofemerging culture that I think is
trickling around around, likegetting really hungry to just
figure it out and do the mostefficient thing possible.
I mean I can sort of like claimbeing really long in the tooth
and saying, like early in mycareer, I can imagine a time
saying like I don't know how todo something or I didn't do
(31:03):
something because I couldn'tfigure it out or I didn't have
the tools to do that.
That's not a, that's not a likean objection anymore in doing
work, and I would encouragecolleges to kind of approach,
approach AI with the same typeof lens, which is like there's
no reason to not do things moreefficiently, and to look at
(31:25):
points of leverage, whether it'sautomating the front door in
the admissions office or keepingstudents successful.
If I even think about divingdown into that question, mike,
around what can be different inthe way we recruit students or
the way we retain students?
I think the admissions use caseis really well known.
(31:45):
Like school to school, schoolswill tell me all the time like
well, we're really unique, werecruit students in really
different ways, and I'm like, haha ha, no, you don't.
And like I can see, like you'resmiling at me right now, which
tells me like you and I haveheard of the same thing.
Like schools will claim likewe're really different, and I'm
(32:05):
like no, no, no, you're not.
That's just not true on theadmissions and recruitment side.
On the student success side,though, schools are really
different, and this is where thedifference of value prop school
to school is actually reallyremarkable, both in terms of who
owns that number and who ownsthat metric, how you report
against it, and then whatprograms and services you
deliver to your students inorder to support that number and
those goals.
So I actually think like somereally interesting innovation is
(32:29):
going to be happening in thestudent success space, because
all of the jobs to be done inadmissions are being handled
right now right Like you reachout to students.
They have an application, theyhave something to do, you have
something to do.
That process is really known.
Student success in general, andstudent service in general, has
been so very limited to thehumans that have been delivering
(32:50):
those programs and services.
We can't even imagine becauseeven if you think about advisors
, they are tasked with acaseload of 500 kids.
How well can you get to know500 college students who are
dealing with a huge range ofcomplex issues as humans?
And we have only limited ourlike, what can we deliver from
(33:14):
the student services side towhat the humans who are staffing
this can deliver, and I thinkit has been really inadequate
and I think that better use oftechnology and AI and agents is
going to be the thing that notonly allows us to solve the
problem but really like startimagining what those jobs are
(33:35):
that we need to do in order tokeep the students successful.
So my argument there is like Idon't even think we've thought
about what those things could be, because we've always been
limited to what the human eventhink.
We've thought about what thosethings could be because we've
always been limited to what thehuman resources that we've had
could deliver for us.
Speaker 1 (33:50):
Completely, completely.
That's such a fascinating pointwe don't know what those end
solutions could possibly be whenyou think about and my bias
here is almost sort of going tolike the tiny bit I know of,
like training, like an AI model,right.
Or like crafting, like a reallygood prompt to start to kind of
(34:10):
get things.
When you think about um, youknow we've got all these sort of
different challenges that aresort of in the space.
We've got, you know, decliningdemographics, shrinking in
different areas, we've got coststhat are way too high, we have
administrative overburden.
We have, like, I think, areally sort of big, massive
(34:31):
institutional challenge, which Ithink you guys are actually
helping self with regards tosort of data lakes that don't
connect to one another.
Right, so you've got this sortof very sort of disparate data
sets.
I think you've got all thesesort of interconnecting pieces
that I think kind of get to thepoint you made earlier, which is
, you know administrativeefficiencies just aren't there
(34:51):
right, like when you think aboutgoing to a place where it's
like you don't know what thesolutions are.
But you've got you know kind ofwhat some of the problems are.
And from your customers, youknow a lot of what those
challenges are.
From the students, you know alot of them.
How do you think about likegetting to those really
interesting solutions and sortof like thinking three to four
steps ahead about what thosecould possibly like be?
(35:14):
What are the data that you'rereally sort of curious of
looking at?
What are the sort of trends inthe sort of broader space that
you're just curious about?
Where's your brain go with thatkind of stuff?
Speaker 2 (35:26):
Yeah, I'm always going to start with like where
we can learn from leadingindicators and I think, pulling
leading indicators versuslagging indicators and this is
like, again, something we talkabout potentially in tech that I
don't know, that we talk aboutthe institution side.
But if we can focus our energyon leading indicators, then the
behaviors that result thenbecome a little bit clearer,
(35:49):
because lagging indicators aresort of like the Gartner
maturity curve of analytics,which is to say, like looking in
the rearview mirror tells yousomething.
It helps you analyze the past.
But actually higher order workis when you can become
predictive and then prescriptiveright to predict what might
happen and then be prescriptiveabout how you might change that.
(36:12):
And I think driving up on thatmaturity curve of analytics is
actually a good way to thinkabout how you might choose
different behaviors andstrategies, whether that's like
a solution that you might buy orhow you might direct the staff.
Speaker 1 (36:29):
Yeah, love that.
I do think that leading versuslagging indicator is just
something that is very sort ofoverlooked.
Yeah, and really sort of likethinking about sort of how do
you differentiate between thosetwo very, very different things?
Speaker 2 (36:43):
Yeah, well, and I think, like you're sort of the
first part of that questionaround, like sort of how you use
AI and how you train AI models,and you know, you and I had
started talking a little bitabout like how institutions are
thinking about AI right now and,really, like all institutions,
when you think about solving it,particularly a technical
problem rightly, you're going togo through this exercise of
(37:04):
like, do we build this or do webuy this?
And I think there's like a caseon AI and sort of the way you
might evaluate like an AI vendoror AI partner or something you
might direct your IT group tobuild for you.
Some of the things that youwant to think about are like how
you train it, how accurate itis and how, like, genuinely
(37:24):
helpful it's going to be, andsort of whether that's around
like providing access in a moredemocratic way and more services
to students, or if it's reallyaround sort of accuracy,
correctness and sort ofalignment with the institutional
tone.
All of that is really importantand some of the things that
we've played out on our side tocreate products that do that.
(37:44):
It's not just one AI model,it's 30.
So when you interact with our AIrecruiter.
You're not just getting likeone LLM.
That's like a yes man to you,right?
It's as if you're likeconsulting a whole room of smart
people.
You know, if the question islike, hey, do you have a good
tennis team?
You know it's like consulting awhole room of smart people,
(38:06):
like, yeah, we have a tennisteam, correct?
Yes, okay, is it good yet Okay,yes, no, okay, okay, our tennis
team might not be good.
Let's, let's actually deliverthis answer and, just like you
would deal with sort of aconference of smart people.
Multiple AI models actuallyprevent poor information, poor
tone, in order to get thatcorrect information and good
(38:28):
tone out to students.
Speaker 1 (38:31):
Yeah, that makes a ton of sense.
You know, I was giggling whenyou referenced the sort of
differentiation piece like forcolleges and universities,
because this is something I seeall the time and you do too, I'm
sure, in your work.
Right, it's like the admissionsoffice that is sort of trying
(38:53):
to sort of pack, or themarketing department that's sort
of packaging this sort ofunique things.
It's like those three to fourpillars and the individualized
attention like the things thatthey're selling are like not
necessarily sort of unique inthe market and usually they have
to sort of like dive deeper toreally like sort of surface that
right when it comes to likeyour own product, if you were
sort of giving the pitch likewhat's unique about you, because
I think I could sort of mayberegurgitate it back to you in a
(39:14):
way that wouldn't be true.
But like, when you think aboutwhat's unique about college fund
, how do you practice that forsomebody?
Speaker 2 (39:26):
It's super easy right now and I see this.
You know all the way down tolike when we do, you know, a
final purchasing process with a,with a president or a provost.
Um, I've sole sourced a lot ofthings for a lot of public
institutions over time.
None of them have been trulyunique.
Right, like I have claimed thata CRM is unique, I've claimed
(39:47):
that a certain service is unique.
Nobody else is doing what we'redoing right now in the space in
terms of innovating an AI agentplatform to automate different
jobs across campus.
An AI agent platform toautomate different jobs across
campus.
That's certainly true in therecruitment space particularly.
So claiming uniqueness rightnow is easy.
But even if it weren't easy,let's say we had lots of
(40:08):
competitors in the space doingthis job.
I think what makes CollegeVinestand out is that we are, at
heart, like a network productwhere, or like a really a
network, because you might notknow this about College by the
Mic.
We have millions of students onour network who are doing their
college journey.
They're accessing tools forfree, they're getting connected
(40:31):
to colleges that they care aboutin ways that feel really good
to them and we are bringingaccess again, like with sort of
democratizing the process inmind.
We're bringing access tostudents who otherwise can't
afford or wouldn't afford areally expensive guidance
counselor to get them into areally elite college.
We have plenty of thosestudents as well.
(40:51):
Those students are reallyactive in the process and we
certainly have many of them onCollege Vine.
But for the students who don'thave access to good guidance
from their high schools from nofault of their own or from their
high schools we can serve themin ways again we can sort of
make that job much moreefficient for them with the help
(41:12):
of AI and really good toolsthat are free for them.
So that's really an importantpart of everyone here on the
network whether it's collegesdoing their work, students
joining to get that goodguidance and being a network
really helps us to smooth everysingle part of it.
So claiming uniqueness is apiece of it.
(41:32):
Being a network overall isanother piece of it.
And then the way I think wetrain our models is really
important, and you should have amindset of training AI Like you
would train a really smartstaff member, where, if you got
a new employee, you would likemarch them through the parade of
HR where you would have themwatch some compliance videos
(41:52):
about, like how to lift withyour legs and how not to like
take bribes at work.
Right Like there's thecompliance.
Speaker 1 (42:00):
HR parade.
Speaker 2 (42:01):
But then, after you got them through that, like what
would you do?
You would teach them a bunch ofthings about how to act as an
agent of your institution.
You'd give them a bunch ofstuff to read and you would have
them set up a bunch ofconversations with smart people
at your institution so that youcould learn not only like what
to say, but how to represent theinstitution.
You are representative of theinstitution in whatever job that
(42:22):
you do.
We really think that that's animportant thing to retain.
As far as training our modelsand having schools work with our
AI models, that that mindsetshould be, you know, really high
order, like as if you weretraining, as if you were
training a really smart staffmember, and the results, really
(42:44):
the results really speak forthemselves on that front.
Speaker 1 (42:47):
Yeah, yeah, I'm so glad you brought the front end
sort of piece of your productwhere you're helping, sort of
shepherd students through thisvery sort of complicated process
, because that's a totallydifferent value proposition
right than the one that you'reoffering to students.
I have to assume that thatprobably comes with its own
challenges right, where you'vegot this challenges and
(43:09):
advantages and opportunitiesright, where you've got a sort
of really sort of diverse andmultifaceted product that is
serving administrators in someway, and that's where you know
the listeners are going to bemore familiar.
But then you've actually gotthe um, the products that are
actually sort of helpingstudents through a very
convoluted like admissionsprocess and what's the?
What are some of like the uxchallenges of like serving those
(43:33):
two different populations anddesigning tools, product
software etc.
That like that fit the verydifferent needs of the two.
And then what are like somecompetitive advantages you have
by, like you know, matchingthose two data sets in ways that
maybe competitors don't.
Speaker 2 (43:52):
Yeah, I would say that it's actually not that
complicated.
When a core tenant of like whatwe do at the company is sort of
genuineness and candor inservice of helping people, right
Like.
If that's the held belief, thenit's actually really easy to
make product decisions aboutwhere we spend our developer
(44:12):
time or what features that werelease right Like, even in
terms of you know, somethingsmall like the way that we
connect students to colleges.
Because if you've got studentson our, on our site doing their
journey, the way that they getconnected to colleges outside of
Collegevine feels really bad Inthat model this is the student
search model that we werestarting to talk about earlier
(44:34):
their name gets bought and soldin ways that they can't control
and in ways that feel really badfor them.
Then the follow on effectsespecially if you're a college
that is trading on this brand oflike we give a shit about you
and we want to get to know you,but you've met them through this
mechanism that feels really bad, like that's a broken system.
(44:56):
So if you take that same smallexample about how students get
connected to colleges and youthink about how to do that
better, you let students controlit.
You let students say yes or noto being connected with a
particular college and becausethey control it, they are much
more willing to share with uscandidly about who they are,
(45:18):
what motivates them, what theirpreferences are, and those
preferences and data points notonly help us understand what
their challenges are and whatproducts we should create for
them, but helps us advise ourcollege partners about what
their population of studentsreally care about.
So there is a genuineness andcandor component that actually
(45:39):
makes those product decisionsreally easy, because then you
get to just drive behaviors outof those sort of that felt
belief or that held belief.
The behaviors and tactics thatyou deploy just feel really
simple, all in service of that.
So I don't know that it'sactually that complicated with
that at the center, because thatis the thing that frees us up
(46:03):
to really then focus on like,okay, what do we innovate on?
Where can we gain efficiencies?
What might we makerecommendations to a college to
actually do?
Speaker 1 (46:11):
Yeah, I love that answer and just letting your I
think values and beliefs reallysort of guide some of your
decision-making there.
You've been super generous withyour I think values and beliefs
really sort of guide some ofyour decision-making there.
You've been super generous withyour time.
Like, what are you excitedabout?
Like there's so much sort ofhappening in this space.
You all are building somepretty incredible things.
Like what gets you personallyexcited?
Speaker 2 (46:32):
I am personally excited for like solving more
stuff for students.
I can remember being a collegestudent and not knowing how to
run, you know, my college searchjourney Once I got to college
being really confused when I hadto go to the registrar and ask
for a transcript, notunderstanding the systems that I
(46:52):
was a part of, and in my day itwas like, you know, going to
talk to like old Marge in theregistrar's office and you had
to go do something in triplicateand then you had to like mail
it here and put it in this slotand like it just seemed like a
lot of sort of like magic that Ididn't really understand how I
was part of it.
And I think that problem stillexists today, where students
(47:14):
actually have a higherexpectation of how they're
treated like as consumers.
Right To like that we shouldknow more about them and treat
them better.
And I think the stakes arehigher for institutions right,
because with more and moreinstitutions closing and being
under like serious financialduress and that's bad for
communities, we have to sort ofdo two things One is like become
(47:36):
more efficient as an industryand also like meet a higher
standard and I actually thinkthat better use of AI with data
underneath it is the key leverto how that's going to happen,
and I can see this in, like youknow, if I think about, if I put
my sort of like student searchhat on as an enrollment marketer
(47:56):
, I'm pretty happy with a schoolthat buys a list of names and
then you market to that list.
You get like a 19% open rate andI think that's pretty good.
When I think about what AI isdoing, right now, it's closer to
a 40% open rate Right, so Itrack that over time and then I
watch the efficiencies thatstart to fall out of that, which
(48:17):
is like, oh, we could run asmaller funnel.
We could actually do this moreefficiently, find better fit
students and create classesabout the things that we want.
Like, another layer of what Ican do on sort of the class
building sides of things is helpyou index on like we are a
college that creates leaders andwe need a really civic minded
(48:41):
class.
That's a job that you can haveAI go do across so much data
that we like cannot hold in ourhuman brains.
Speaker 1 (48:50):
Oh yeah.
Speaker 2 (48:50):
So I think overall, like the solving of that higher
standard that students have andmaking it more efficient at the
same time and sort of that likebig orchestration of all of the
things that need to happen tomake those two things true, are
what I'm just like super excitedto be a part of, because in my
journey I got to see from Margein the registrar's office and
(49:12):
the frigging triplicate piecesof paper to like what we can do
today and it's just thrilling towatch that continue.
Speaker 1 (49:21):
Couldn't agree more.
I think that's a fabulous placeto end it.
But, yeah, thank you for thetime here, emily, and yeah,
please let people know how theycan get in touch with you.
Speaker 2 (49:32):
Yeah, so you can find me on collegevinecom, on
LinkedIn, at Emily Smith, andyou can also check out.
I also have a podcast, which isreally similar to this one.
So if you like the topics thatMike talks about, I too have a
podcast.
It's called the Weekly VineDown.
You can find us atthevinedowncom and I hang out
(49:53):
there putting out a new showabout every week.
Speaker 1 (49:56):
Awesome.
Well, this has been awesome.
Thanks again.
Speaker 2 (49:59):
Thanks for having me, Mike.
I am psyched to be here with Emily.
Emily welcome.
Speaker 2 (00:04):
Thanks for having me, Mike.
I'm really, really happy to behere with you.
Speaker 1 (00:08):
Same here.
You've got such a fascinatingbackground and maybe just to
sort of square the work thatyou're currently doing and the
problems that you're solving atCollegeFind can you just walk us
through your career trajectoryand how you went from serving
higher education with startupsto then working for some
enterprise solutions and thenyour current work at Collegevine
(00:30):
how that all come to pass.
Speaker 2 (00:33):
Yeah, easy question, right to be asked about yourself
and how you got here.
It's all 2020 in the rear view,but it started for me as a
senior in college and this willmaybe like warm your career
services heart, mike, but like Idon't know why, it didn't occur
to me in the second semester ofmy senior year that like I
would have to leave college andget a job and that somehow felt
(00:55):
surprising.
And I I'm a ski bum at heartand I was like, okay, well, I'm
just gonna, you know, ride offinto the sunset and and ski my
little brains out.
And then I realized no, youactually need to make money and
contribute to earth.
And at that time I had beenworking with my career services
office, who encouraged mebecause I had been a tour guide
in the admissions office at myalma mater to talk to these
(01:16):
consultants who were coming in,who kind of wanted a generalist.
And I took the meeting and I'mso glad I did, because that set
me off working for a consultinggroup doing enrollment
consulting and services rightout of college.
And I got pulled into roomswith university presidents and
provosts and vice presidents ofenrollment at 22 years old,
being asked what I thought aboutthings, which was a really
(01:38):
powerful way to feel reallyimportant early in my career.
So I got pulled along like thelittle sister of these serious
enrollment consultants, fixingserious enrollment problems, and
after that I moved on to reallythe software product side of
things.
I was the first employee at acompany called Fire Engine Red
where we really helped to inventemail marketing for colleges
(02:00):
and universities and did a lotof software products aside from
the CRM on the enrollment sideof things.
We also did a lot of sort oflike software products aside
from the CRM on the enrollmentside of things.
We also did a lot of studentsearch and the student search
when I and by that I mean likepurchasing names and marketing
to them, doing the enrollmentmarketing via purchase names of
test takers and and other otherpeople like that, and in that
era that motion like workspretty well and it started to
(02:24):
work less and less well overtime, um, and I really enjoy
being part of the small, smallbut growing thing that companies
do.
Right, I was the first employeeof a company so I got to do all
the jobs, um, and as I movedthrough my career, working, you
know, at another company doingCRM for admissions and student
success.
Again.
I joined as a smaller companywas part of growing that company
(02:46):
to be bigger and bigger and Iseem to keep going back to the
small scrappy and being therefor the really exciting growth
and I really love that because,as slow as we observe, colleges
moving companies can move reallyfreaking fast and it's really
really fun to like test stuff,fail and iterate in that cycle
and to me there's this throughline of like really figuring out
(03:09):
how to get a pulse on whatstudents serve students really
well and that's changed a lotover time and being a part of
feeling like a fundamental fixerof those problems is just
really important to like my, mypersonal mission.
Speaker 1 (03:27):
That is so cool and I appreciate you contextualizing,
ensuring all those thosedifferent pieces and those those
three lines like how do yousort of sit at the intersection
of, like those two verydifferent things?
Right, you're serving this likevery traditional industry that
tends to sort of move slow,right, where iteration is not
necessarily sort of baked intothe culture where they're
(03:50):
thinking in timelines that aresort of decades long, but you
juxtaposition that with the typeof work that you're doing,
where it's like it's quick, it'siterative and it's like, and
it's serving all these spacesLike.
how do you hold those twoextremes and find that happy
middle ground between the two?
Speaker 2 (04:07):
Colleges are not startups.
If you've ever worked for, likeon the corporate side of things
we think about, you know bigcompanies that do things like
really.
I guess a rude reading ofworking for a really big company
is like it might not beapparent.
If you didn't show up for workfor like a week, two weeks in a
row, like your impact would notnecessarily be felt.
(04:28):
And colleges are a little bitlike that in that they are
really established businessesthat we expect to be open and
available for jobs, for theproducts that they provide, a
community, and in startupsoftware we have to be so
scrappy and everybody'scontribution, like every
individual employee in a startupcompany, their impact really,
(04:51):
really matters.
So to bring some of that mindsetinto higher education, which is
like bigger, a little slowerand a little more stayed, feels
really exciting on theinstitutional side.
I think it lands really well.
I also think there's some sortof leverage and power that you
just get as the outsider right,like you might be saying
something as a consultant thatyour cabinet had been sharing
(05:12):
with you, that you're justhearing it differently from a
consultant, and I understandthat there's power in that.
And the third thing that Iwould mention is really around
like the advantages that I havesitting in my seat, peering
through the windows like this atlike hundreds of institutions.
Folks I work with on theinstitutional side.
They know their politics reallywell.
They know the ins and outs oflike how things are politic or
(05:36):
clicked through in their owncampuses.
But I get this advantage oflike bringing in this broader
outside perspective, havingworked with hundreds and
hundreds of colleges over time.
So my observations, whilethey're maybe a little bit
broader than you would give fromthe inside, get to be
transformational just because Iget to be the outsider.
Speaker 1 (05:57):
Yeah, super cool, I love that answer.
Can you walk me through alittle bit about like college
run specifically?
Um you all have leaned into,which I think is just so awesome
?
Um focused on the sort of aispace and the ai integrations
and supporting particularly sortof enrollment.
(06:18):
Um you know how.
How have you sort of like what?
What's collegeVine's journeybeen like?
And then what's been yourunique contribution?
Because you're sort of bringing, I think, a longer narrative
and historic lens about likedeeply understanding CRMs and
seeing that sort of originalmigration to email and all these
(06:40):
sort of like bigger techchanges to serve a traditional
industry.
What's it been like to sort ofbe on that front line of AI and
what's been your own sort oflike product journey at
CollegeVine to land in where youare now?
Speaker 2 (06:53):
Yeah, so I've been at CollegeVine the last three
years.
The company I'm not a founder,the company is many years older
than that and has been through acouple of different iterations,
and the way I would describewhat Collegevine contributes to
the space now is that we're aplatform for building and
deploying AI agents atinstitutions, and that is a sort
of fifth order thing that we'vegotten to over time, because
(07:15):
there are a bunch of otherthings that we do in service of
helping students and familiesget the best guidance possible
at every stage of life, whetherthat's the transition from high
school to college or beingsuccessful as a college student
or transitioning from college toa job, and there is a limit to
what humans can do, whether it'sat the sort of guidance layer
(07:38):
at all of those stages or at theadministration layer of all of
those different phases for astudent.
And now, really, as an AIcompany, we have created kind of
autonomous AI agents that goout into the world and do work
with constituents, and that'slike things that you would if
you were a college president orprovost listening to this and
(07:59):
you're thinking.
You know the ways in which youlimit the requirements that you
put on your staff or strategiesthat you create to the humans
that can deliver them.
We don't have to do thatanymore.
We don't have to put thoselimits on anymore.
We can really dream much biggerthan we have been able to in
(08:19):
the past to think about what AIcan do as a really strong point
of leverage for automatingdifferent jobs at the
administration layer of ourinstitutions.
And that's the problem thatCollegeVine is really solving
right now in the college space,while bringing along students
and families in a way that feelslike it delivers really great
(08:42):
access to guidance andconnections across that whole
landscape.
Speaker 1 (08:46):
That's cool.
Can you walk me through like afew of your favorite like recent
use cases around, like howyou're solving those sort of
massive operational pain points?
Right, because it strikes me asa you have such a unique like
product and engineeringchallenge just given, like the
sort of the data, the datalayers that you have, the
challenges with integration thatyou have with all the different
(09:10):
third-party pieces, which Ithink you've had really smart
APIs and connections to solvefor those.
But can you walk me through afew use cases and then what the
felt experience is on the sideof the user or the admin?
That's really doing it.
Speaker 2 (09:28):
Yeah for sure.
So we've got several sort ofout-of-the-box AI agents and
probably the easiest one toconceive of how it does its job
is the AI recruiter.
This is something that we arelive with at several hundred
colleges, live with at severalhundred colleges, and the AI
recruiter essentially is anautonomous agent that delivers
(09:49):
recruitment activities for theadmissions office, whether
that's pairing one to one withstudents to like you would want
a counseling staff.
If you could afford it, youwould hire lots of admissions
counselors to pair one-on-onewith students to give them an
individual journey and treat,like you, mike as an individual
student, versus like ah, mike isa guy from Wisconsin who's,
(10:10):
like, interested in engineering,so we're going to bucket him
with all the other guys from outof state who are interested in
engineering.
It's inherently a differentexperience when you're working
one-to-one with a really smartadvisor, so that's one job that
the AI recruiter doesno-transcript On day 13,.
(10:52):
We send this email and then wehave to go write all those
things and create creative for.
It Turns out AI is really goodat that job and, in fact, better
than we are, which I have somefeelings about, having spent all
those logged, all those hoursdoing that job.
But AI is better than I am atit.
And the third job that therecruiter does right now is
processes, transcripts andapplications.
(11:14):
So that's again something thathumans have done for a long time
to grid highest math or well,we take this GPA and we
calculate it in this way andthat's a really expensive
operational problem for collegesto solve.
That the AI recruiter does.
So that's how to conceive ofone of the AI agents.
But the AI agents show upacross this whole student
success journey, being anadvisor to students, a concierge
(11:36):
, to either find stuff out likehey, mike, your housing form is
done.
Like why haven't you done it?
Or hey, mike, how are youfeeling?
Like have you made any friendsRight To really get at early
indicators of how students aredoing in their current student
journey.
And the applications continueto extend through the ways in
(11:56):
which we would interact withalumni or build networks for
career services.
But at the administration layer, we're thinking about all of
the different work that we canautomate in order to do it
better, faster and bigger.
Speaker 1 (12:18):
I absolutely love that answer of the felt sense of
what it's like to sort of be anadmin working with this tool or
a student that's really sort ofon the receiving end and doing
the different pieces.
It's so fascinating what youmentioned around that roadmap
and one institutional challengeI see like again and again is
like with turnover in the spacelike being relatively high and
(12:41):
documentation on process andthings like that, not depending
on institutions sort of beingbaked hard into the culture.
You have a variance around whatthat's going to look like from
year to year, depending on who'ssort of connected and different
pieces there.
That's just going to sort ofchange and iterate, not
(13:04):
necessarily in good ways, andyou also have this sort of lack
of data capture and processcapture and all these sort of
connected pieces there.
What are the feedback loops forthat on your end to sort of
ensure, like you know,communications are sort of
working well, like what are thethings that you're looking at
institution basis and how areyou empowering partners to sort
(13:27):
of like?
do some of the assessment piecesconnected to all that.
Speaker 2 (13:32):
Is the question like, is part of that question, like,
like, how do we know it'sworking?
Like, how do we measure it'sworking?
Is that sort of at the heart ofit.
Yeah, exactly, such a goodquestion, right, because, like,
in this industry turns out funfact, we're not like shaking the
money tree and it's not likeraining down dollar bills for us
to like try stuff that doesn'twork, right, like we're at a
(13:52):
critical point on making thebusiness of the institutions
work really well.
So things that work are reallyimportant and the ways that we
look at sort of measuringsuccess is like imagine your
highest order expectation foryour best possible staff member.
And if, mike, I hired you at mycollege to work in admissions
(14:13):
for me, I would want you everyweek, every day, if I wanted you
to, to be like handing me areport card of how you were
doing.
I would want to know how manystudents you were talking to.
I would want to know how manytimes you messed up.
I would want to know how manytypes of different questions you
were getting.
I would want to take that dataand be able to synthesize it to
the faculty and creators ofprograms at our institution to
(14:35):
be like yo, we're getting askeda lot for this type of program
that we don't offer.
What are we going to do aboutit?
Like, is that something we'regoing to change the business to
accommodate?
Or look, we can very easily seethat this is no longer a
program that is offering anycreating any real genuine
interest or excitement with ourpopulation.
Do we need to change theprogram?
(14:56):
Is it profitable for us to keepthe program?
So anything that you can thinkof, sort of your highest order
expectation of what a humanstaff member would deliver you
to provide you real leverage inleading the institution, are the
types of things that the AIagents sort of bring you and
help you measure, because it'sreally like that leverage that's
(15:17):
going to allow you to be makinggreat strategic decisions for
the institution.
Speaker 1 (15:23):
Yeah, very cool, very cool.
What are some other use casesthat you're excited about?
I mean, the AI Recruiter issomething that I'm just talking
with a lot of people People areexcited about.
The feedback has been reallygood.
It seems like it's sort ofrelatively plug and play and,
despite the sort of complexityof all the interconnections that
you need to make something likethat work, it seems like it's
(15:44):
working really effectively.
What are some other use casesthat you're pumped about?
Speaker 2 (15:48):
Yeah, so the stuff that we're creating now, like
imagine, if you so I'm going toask you to like go back in your
time machine to the time thatyou were leading career services
offices, mike, and like imagineif you were sitting with a kid,
like human, you were sittingwith a student and the student
was, you know, about to embarkon their career search.
(16:09):
And this is the point ofcollege right To create people
who can go be good citizens ofearth and contribute to our
societies and cultures.
Like that's the point, and you,as the career services officer,
are trying to make that happenin the best possible way, like
you can.
Officer, are trying to makethat happen in the best possible
way, like you can.
And imagine if I gave you atool that, like allowed you to
(16:29):
connect this student.
Let's say the student was,let's go with engineering.
Let's say the student was anengineer and was really feeling
nervous about, like, what lifelooks like as an engineer.
And what if AI could help younot only identify like sure
tabular data could do this todaybut identify who from your
alumni network worked as anengineer and was open to
(16:50):
mentoring.
Like you could get that withspreadsheets and joins in a good
enough database today?
Sure, but what if AI was thenable to go facilitate that
introduction?
Because the friction is in theintroduction making in that use
case, like, really specifically,the friction is you write down
the name of a person and say tothe student like hey, look this
person up, go build arelationship with this person.
(17:11):
That is freaking terrifying fora student.
But imagine if you had an AIagent to not only identify the
best fit alum who could go dothis job for you, also indicate
their willingness to be a mentorand also then kick off this
relationship and facilitate thisrelationship between two humans
.
That is a really high leveragemove you can make as a career
(17:35):
services person.
And there are a billion usecases like that that keep the
human fundamental relationshipsor transformational relationship
building at play, which issuper important to colleges, but
adds a techification on top ofit to make it go way faster or
better.
Or if you were a professor andyou had an incoming suite of
(17:56):
students in your senior seminarand you had your AI agent or
your AI assistant telling you oh, we've got these 20 students
you really need to watch out forso-and-so.
She's really struggling withsome other classes.
So be aware of this.
This other student is going tobe a great leader for you.
This other student hassomething going on in their home
life that you should be awareof.
This other student is headingtowards a PhD in your program.
(18:18):
That would be a really helpfulhigh-leverage thing that you
could get very quickly as aprofessor.
That would fundamentally makethe way you did your job better,
facilitated by better data andtech.
Right Like sure you could wadethrough the data that we have
today to make sense of that.
You're never going torealistically do that because
it's way too high friction.
Speaker 1 (18:39):
Totally yeah, and I love the career services example
.
I deployed PeopleGrowth on acouple of campuses and it was so
fascinating to just look at thedata sets around goes to
somebody like but you have, evenif, like you execute really
(19:08):
really well these tiny, tinysteps, you lose people along the
way.
So any way that you can sort ofintegrate that like very sort of
seamlessly, like your chancesof like the person actually sort
of doing the thing, which islike making that initial contact
.
What we figured out right away,like very quick, was like if we
got the person to send onemessage, their chances of
(19:29):
actually coming back were likeexponentially higher and sort of
using the tool where we werereally high.
But like there's so many manysteps between like the person,
like getting on the platform andthen actually finding that
first person that you walk likethat you connect with that like
if you can close that gap, likethat is a massive lever in terms
(19:50):
of like the person's careertrajectory that you don't really
sort of think of.
But it's like a bunch of ministeps consolidated.
If you can solve those painpoints, it's it's huge.
So it's really cool to see thatyou're doing this in like a lot
of different ways.
Speaker 2 (20:05):
Yeah Well, one of the things that we talk about, like
on the sort of corporate sideof things I'll sort of like I'll
bring a term in from SaaSsoftware that might be
interesting to your audiencewhich is like time to first
value, and essentially, thequicker you can bring your
consumer time to first value andthat's short, as short as
possible, the better, like thebetter life of the customer,
(20:27):
whether it's, you know, revenueor success or whatever.
We're measuring over time andin your instance of like if we
could just get them to send thatfirst email, or, on the
recruitment side, if we can havethat first interaction, be
really positive.
You're right, it creates lessfriction in the rest of that
chapter of the journey.
But that time to first valuething is really interesting
thing to measure across yourconstituents to watch that
(20:51):
friction start to dissipate.
Speaker 1 (20:53):
Yeah, that's really cool.
I've not heard it phrasedexactly like that, but I love
that.
Time to first value as a metric.
I'm just curious is theresimple ways to measure that?
Speaker 2 (21:06):
I'm sure it depends on the use case.
Yeah, it super depends on theuse case, and this is something
that we talk about both in theproduct side of tech companies
as well as the client successside of tech companies, and it's
basically whatever you thinkthe first valuable moment is,
and usually it's whatever feelslike that aha moment.
So on the admission side ofthings we might be really sort
(21:29):
of hard and fast on like well,it's the application, right,
that's the thing that allows usto really gauge interest.
Things in admissions andrecruitment and enrollment are
actually way easier for us tomeasure than other use cases
across the administration layerof an institution.
But it could be something likesentiment.
It could be something like youknow, gartner does a really
interesting survey on a set ofquestions that like allows you
(21:53):
to measure the success of youremployees and one of the
questions is like have you madea best friend at work?
And you ask adults if they'vemade a best friend and they're
like no.
I'm a 45 year old man, like Idon't need to make a best friend
, but really like those types ofthings like a student, have you
made a best friend yet?
You know, like have you joineda club yet?
Are really strong indicators ofhow a student is viewing their
(22:16):
experience at your institution.
So you kind of get to call it.
But when you call it then justlike anything else, we measure,
like the point would be then tomeasure consistency, learn,
iterate, do it again, love that,love that.
Speaker 1 (22:34):
So a question that I can just think of the sort of
audience asking because I knowthis is sort of differs wildly
sort of depending on the usecases and different pieces.
But how do you build your AIagents?
Right, like because you know Ithink the basics that you know
you hear pretty consistentlyfrom AI engineers are
essentially, you're sort ofyou're feeding it a lot of data
(22:56):
and you're doing sort ofsystematic training to sort of
get it to sort of perform theways you want giving feedback
and all those kinds of pieces.
But what's that look like foryou?
Because you have such a varietyof use cases, you're, I think,
increasingly sort of gettinginto this more sort of
customized space and havingsolutions that are sort of both
out of the box but also sort oftweakable.
(23:17):
So what's like the backend ofMagic look like for really sort
of making these things work inthe way that you want them to
work?
Speaker 2 (23:25):
Yeah, that's a good question.
And the other thing, the otherelement that I'll add to it is
colleges do a lot of feelingaround in the space of like we
know we need a thing, do webuild it or do we buy it, and
there's sort of that element,too, which is like, can we do it
ourselves?
And I think what's reallyinteresting with AI is that, you
(23:47):
know, very occasionally I'lltalk to somebody about you know,
the AI recruiter, for example,and they'll just be like, well,
why couldn't we just chat GPT,the AI that you go out and ask
something of is reacting to arequest in a transactional way,
thing of, is reacting to arequest in a transactional way?
(24:08):
In 2025, we are heading into aplace of agentic AI that is more
autonomous and proactive, whereyou give it a whole job.
Yo, ai, recruit the class.
Hey, ai, take all of thecaseload of all of our 40,000
students and make themsuccessful.
Speaker 1 (24:23):
I'm so curious about this with you all because you
guys are doing some really sortof interesting work at the
intersection of building reallycool bots that are doing amazing
things.
What are some unexpectedchallenges that you've faced in
terms of sort of deploying andopening up your range of
products and things that you'redoing, and then how have you
(24:45):
worked to overcome those?
Speaker 2 (24:47):
Yeah, I think a couple of things.
One I think most of thediscourse about AI in higher
education right now is on theteaching and learning side, and
I think that's a debate we needto solve, and solve quickly, so
that's sort of thing.
One, and by that I mean, likewe can't spend time screwing
around with, likeinstitutionally, how we feel
(25:07):
about the use of AI in theclassrooms, Because if you start
to indicate, like a less thanclear stance on this, you will
start to create ripples for yourstudents.
You will start to then havethose ripples become waves,
which will make students notfeel really confident in their
participation of your brand asan institution.
(25:28):
And I guess I care more aboutmaking a stand on this than what
the stand is, Though, of course, I think the stand is look, if
we're trying to createespecially young people who are
going to participate in theworld in meaningful ways and get
jobs, then one of the thingsthat we need to do is to train
(25:49):
them how to use AI well.
It's a job skill.
So of course, I have a verystrong opinion about how
institutions come down on this,but I actually care more that
institutions have a reallystrong stance on it.
I think, once we can let someof that air out of the room on
the focus that we spend on AIfor teaching and learning, we
need to quickly turn ourattention to what AI can solve
(26:13):
at the administration layer,Because I think about
institutionally if you correctfor hospital revenue and if you
correct for some of the sort ofenrollment, demographic, student
search cliffs, it's about a $93billion deficit on the
administration layer thatschools and what jobs we would
want to have AI automate.
(26:45):
Then I think it really frees usup to think about what
challenges and strategies thatwe might want to solve.
And one of the surprises that Ihave is the difference between
leadership at an institution.
Some of them are thinking like,okay, yes, actually, our human
costs, our cost of the humanstaff, are actually way too
great.
We need to pare that down andhave AI take over the work.
(27:07):
And then the other camp, whichis no, protect humans and jobs
at all costs.
And I sort of say to them like,okay, great, but let's think
about the ways in which using AIbetter at the administration
layer can give your existingstaff a promotion, and by that I
mean like if you use AI well,you are essentially giving
yourself a promotion.
(27:28):
So if you used to be the personwho compiled data, AI should do
that for you and you can be nowthe person that analyzes the
data right.
There's like a leveling up thatwe can offer to our staffs.
So one of the surprises I haveis sort of like how much time
we're spending on this teachingand learning debate, which I
think needs to end.
And the second is around likethis divided road on should we
(27:53):
automate people out of jobs orshould we automate enough jobs
so that people can be leveled upin their jobs, and those feel
like really fundamental debatesthat feel important to resolve
in the next chapter.
Speaker 1 (28:07):
Yeah, I think they're definitely sort of huge debates
and I completely understand thethinking through the second or
third order effects of like hey,if we're downsizing our staffs
in this way, this way, this wayor this way, you know we're
potentially looking at a verysort of different workforce and
I totally get that angle of.
(28:28):
At the same time, I meanespecially at like a private
institutions or I think youcould almost say it's sort of
non-community colleges, like thecosts to the end student is
paying are so high, right, and Ithink we are in an environment
where good, bad or indifferent,the signaling from the federal
government at this point is youknow, there's not more federal
(28:51):
dollars that are going to likean influx, that are sort of
going to be coming.
I think we've also seen thefolly of a lot of institutions
really relying on their donorbase when we know those dollars
from a outside of like a smallpercentage are going down,
giving rates are sort of goingdown.
So I think that, like I wouldargue, the writing is on the
(29:14):
wall in a lot of ways that likehey, we don't have this influx
of cash that are coming toundercut the cost of like really
sort of growing administrations.
How do we do this work moreeffectively?
Um, and I think you all solvelike a really sort of
interesting thing there.
Um, for for you like what, whatare like some of the like when
(29:37):
you're looking like two, threesteps ahead, what are like some
of the really cool things thatyou think your tech could sort
of jump in and solve that likepotentially.
I don't want to say likeeliminate jobs, but just
heighten the experience of likehow a person sort of sitting in
this sort of like adminfunctional role could maybe sort
of do the work differently.
Speaker 2 (29:58):
Well, I think about.
I think about that mindsetright, that like more money is
not coming, and I think thatbreeds innovation.
And we can choose to beflattened by that and feel
really scared about what thatmight mean, or we can choose to
let that make us curious andhungry to really like flex and
(30:19):
do different things.
And one of the things that Isee happening in the world of
work that I don't know ishappening in colleges yet is a
culture of like efficiency andfiguring it out right, like in
our jobs, and I think about theways that we, you know, like
create our work culture atCollegevine, which we've done
really, really intentionally,just in terms of how we do work.
(30:42):
There's a really new sort ofemerging culture that I think is
trickling around around, likegetting really hungry to just
figure it out and do the mostefficient thing possible.
I mean I can sort of like claimbeing really long in the tooth
and saying, like early in mycareer, I can imagine a time
saying like I don't know how todo something or I didn't do
(31:03):
something because I couldn'tfigure it out or I didn't have
the tools to do that.
That's not a, that's not a likean objection anymore in doing
work, and I would encouragecolleges to kind of approach,
approach AI with the same typeof lens, which is like there's
no reason to not do things moreefficiently, and to look at
(31:25):
points of leverage, whether it'sautomating the front door in
the admissions office or keepingstudents successful.
If I even think about divingdown into that question, mike,
around what can be different inthe way we recruit students or
the way we retain students?
I think the admissions use caseis really well known.
(31:45):
Like school to school, schoolswill tell me all the time like
well, we're really unique, werecruit students in really
different ways, and I'm like, haha ha, no, you don't.
And like I can see, like you'resmiling at me right now, which
tells me like you and I haveheard of the same thing.
Like schools will claim likewe're really different, and I'm
(32:05):
like no, no, no, you're not.
That's just not true on theadmissions and recruitment side.
On the student success side,though, schools are really
different, and this is where thedifference of value prop school
to school is actually reallyremarkable, both in terms of who
owns that number and who ownsthat metric, how you report
against it, and then whatprograms and services you
deliver to your students inorder to support that number and
those goals.
So I actually think like somereally interesting innovation is
(32:29):
going to be happening in thestudent success space, because
all of the jobs to be done inadmissions are being handled
right now right Like you reachout to students.
They have an application, theyhave something to do, you have
something to do.
That process is really known.
Student success in general, andstudent service in general, has
been so very limited to thehumans that have been delivering
(32:50):
those programs and services.
We can't even imagine becauseeven if you think about advisors
, they are tasked with acaseload of 500 kids.
How well can you get to know500 college students who are
dealing with a huge range ofcomplex issues as humans?
And we have only limited ourlike, what can we deliver from
(33:14):
the student services side towhat the humans who are staffing
this can deliver, and I thinkit has been really inadequate
and I think that better use oftechnology and AI and agents is
going to be the thing that notonly allows us to solve the
problem but really like startimagining what those jobs are
(33:35):
that we need to do in order tokeep the students successful.
So my argument there is like Idon't even think we've thought
about what those things could be, because we've always been
limited to what the human eventhink.
We've thought about what thosethings could be because we've
always been limited to what thehuman resources that we've had
could deliver for us.
Speaker 1 (33:50):
Completely, completely.
That's such a fascinating pointwe don't know what those end
solutions could possibly be whenyou think about and my bias
here is almost sort of going tolike the tiny bit I know of,
like training, like an AI model,right.
Or like crafting, like a reallygood prompt to start to kind of
(34:10):
get things.
When you think about um, youknow we've got all these sort of
different challenges that aresort of in the space.
We've got, you know, decliningdemographics, shrinking in
different areas, we've got coststhat are way too high, we have
administrative overburden.
We have, like, I think, areally sort of big, massive
(34:31):
institutional challenge, which Ithink you guys are actually
helping self with regards tosort of data lakes that don't
connect to one another.
Right, so you've got this sortof very sort of disparate data
sets.
I think you've got all thesesort of interconnecting pieces
that I think kind of get to thepoint you made earlier, which is
, you know administrativeefficiencies just aren't there
(34:51):
right, like when you think aboutgoing to a place where it's
like you don't know what thesolutions are.
But you've got you know kind ofwhat some of the problems are.
And from your customers, youknow a lot of what those
challenges are.
From the students, you know alot of them.
How do you think about likegetting to those really
interesting solutions and sortof like thinking three to four
steps ahead about what thosecould possibly like be?
(35:14):
What are the data that you'rereally sort of curious of
looking at?
What are the sort of trends inthe sort of broader space that
you're just curious about?
Where's your brain go with thatkind of stuff?
Speaker 2 (35:26):
Yeah, I'm always going to start with like where
we can learn from leadingindicators and I think, pulling
leading indicators versuslagging indicators and this is
like, again, something we talkabout potentially in tech that I
don't know, that we talk aboutthe institution side.
But if we can focus our energyon leading indicators, then the
behaviors that result thenbecome a little bit clearer,
(35:49):
because lagging indicators aresort of like the Gartner
maturity curve of analytics,which is to say, like looking in
the rearview mirror tells yousomething.
It helps you analyze the past.
But actually higher order workis when you can become
predictive and then prescriptiveright to predict what might
happen and then be prescriptiveabout how you might change that.
(36:12):
And I think driving up on thatmaturity curve of analytics is
actually a good way to thinkabout how you might choose
different behaviors andstrategies, whether that's like
a solution that you might buy orhow you might direct the staff.
Speaker 1 (36:29):
Yeah, love that.
I do think that leading versuslagging indicator is just
something that is very sort ofoverlooked.
Yeah, and really sort of likethinking about sort of how do
you differentiate between thosetwo very, very different things?
Speaker 2 (36:43):
Yeah, well, and I think, like you're sort of the
first part of that questionaround, like sort of how you use
AI and how you train AI models,and you know, you and I had
started talking a little bitabout like how institutions are
thinking about AI right now and,really, like all institutions,
when you think about solving it,particularly a technical
problem rightly, you're going togo through this exercise of
(37:04):
like, do we build this or do webuy this?
And I think there's like a caseon AI and sort of the way you
might evaluate like an AI vendoror AI partner or something you
might direct your IT group tobuild for you.
Some of the things that youwant to think about are like how
you train it, how accurate itis and how, like, genuinely
(37:24):
helpful it's going to be, andsort of whether that's around
like providing access in a moredemocratic way and more services
to students, or if it's reallyaround sort of accuracy,
correctness and sort ofalignment with the institutional
tone.
All of that is really importantand some of the things that
we've played out on our side tocreate products that do that.
(37:44):
It's not just one AI model,it's 30.
So when you interact with our AIrecruiter.
You're not just getting likeone LLM.
That's like a yes man to you,right?
It's as if you're likeconsulting a whole room of smart
people.
You know, if the question islike, hey, do you have a good
tennis team?
You know it's like consulting awhole room of smart people,
(38:06):
like, yeah, we have a tennisteam, correct?
Yes, okay, is it good yet Okay,yes, no, okay, okay, our tennis
team might not be good.
Let's, let's actually deliverthis answer and, just like you
would deal with sort of aconference of smart people.
Multiple AI models actuallyprevent poor information, poor
tone, in order to get thatcorrect information and good
(38:28):
tone out to students.
Speaker 1 (38:31):
Yeah, that makes a ton of sense.
You know, I was giggling whenyou referenced the sort of
differentiation piece like forcolleges and universities,
because this is something I seeall the time and you do too, I'm
sure, in your work.
Right, it's like the admissionsoffice that is sort of trying
(38:53):
to sort of pack, or themarketing department that's sort
of packaging this sort ofunique things.
It's like those three to fourpillars and the individualized
attention like the things thatthey're selling are like not
necessarily sort of unique inthe market and usually they have
to sort of like dive deeper toreally like sort of surface that
right when it comes to likeyour own product, if you were
sort of giving the pitch likewhat's unique about you, because
I think I could sort of mayberegurgitate it back to you in a
(39:14):
way that wouldn't be true.
But like, when you think aboutwhat's unique about college fund
, how do you practice that forsomebody?
Speaker 2 (39:26):
It's super easy right now and I see this.
You know all the way down tolike when we do, you know, a
final purchasing process with a,with a president or a provost.
Um, I've sole sourced a lot ofthings for a lot of public
institutions over time.
None of them have been trulyunique.
Right, like I have claimed thata CRM is unique, I've claimed
(39:47):
that a certain service is unique.
Nobody else is doing what we'redoing right now in the space in
terms of innovating an AI agentplatform to automate different
jobs across campus.
An AI agent platform toautomate different jobs across
campus.
That's certainly true in therecruitment space particularly.
So claiming uniqueness rightnow is easy.
But even if it weren't easy,let's say we had lots of
(40:08):
competitors in the space doingthis job.
I think what makes CollegeVinestand out is that we are, at
heart, like a network productwhere, or like a really a
network, because you might notknow this about College by the
Mic.
We have millions of students onour network who are doing their
college journey.
They're accessing tools forfree, they're getting connected
(40:31):
to colleges that they care aboutin ways that feel really good
to them and we are bringingaccess again, like with sort of
democratizing the process inmind.
We're bringing access tostudents who otherwise can't
afford or wouldn't afford areally expensive guidance
counselor to get them into areally elite college.
We have plenty of thosestudents as well.
(40:51):
Those students are reallyactive in the process and we
certainly have many of them onCollege Vine.
But for the students who don'thave access to good guidance
from their high schools from nofault of their own or from their
high schools we can serve themin ways again we can sort of
make that job much moreefficient for them with the help
(41:12):
of AI and really good toolsthat are free for them.
So that's really an importantpart of everyone here on the
network whether it's collegesdoing their work, students
joining to get that goodguidance and being a network
really helps us to smooth everysingle part of it.
So claiming uniqueness is apiece of it.
(41:32):
Being a network overall isanother piece of it.
And then the way I think wetrain our models is really
important, and you should have amindset of training AI Like you
would train a really smartstaff member, where, if you got
a new employee, you would likemarch them through the parade of
HR where you would have themwatch some compliance videos
(41:52):
about, like how to lift withyour legs and how not to like
take bribes at work.
Right Like there's thecompliance.
Speaker 1 (42:00):
HR parade.
Speaker 2 (42:01):
But then, after you got them through that, like what
would you do?
You would teach them a bunch ofthings about how to act as an
agent of your institution.
You'd give them a bunch ofstuff to read and you would have
them set up a bunch ofconversations with smart people
at your institution so that youcould learn not only like what
to say, but how to represent theinstitution.
You are representative of theinstitution in whatever job that
(42:22):
you do.
We really think that that's animportant thing to retain.
As far as training our modelsand having schools work with our
AI models, that that mindsetshould be, you know, really high
order, like as if you weretraining, as if you were
training a really smart staffmember, and the results, really
(42:44):
the results really speak forthemselves on that front.
Speaker 1 (42:47):
Yeah, yeah, I'm so glad you brought the front end
sort of piece of your productwhere you're helping, sort of
shepherd students through thisvery sort of complicated process
, because that's a totallydifferent value proposition
right than the one that you'reoffering to students.
I have to assume that thatprobably comes with its own
challenges right, where you'vegot this challenges and
(43:09):
advantages and opportunitiesright, where you've got a sort
of really sort of diverse andmultifaceted product that is
serving administrators in someway, and that's where you know
the listeners are going to bemore familiar.
But then you've actually gotthe um, the products that are
actually sort of helpingstudents through a very
convoluted like admissionsprocess and what's the?
What are some of like the uxchallenges of like serving those
(43:33):
two different populations anddesigning tools, product
software etc.
That like that fit the verydifferent needs of the two.
And then what are like somecompetitive advantages you have
by, like you know, matchingthose two data sets in ways that
maybe competitors don't.
Speaker 2 (43:52):
Yeah, I would say that it's actually not that
complicated.
When a core tenant of like whatwe do at the company is sort of
genuineness and candor inservice of helping people, right
Like.
If that's the held belief, thenit's actually really easy to
make product decisions aboutwhere we spend our developer
(44:12):
time or what features that werelease right Like, even in
terms of you know, somethingsmall like the way that we
connect students to colleges.
Because if you've got studentson our, on our site doing their
journey, the way that they getconnected to colleges outside of
Collegevine feels really bad Inthat model this is the student
search model that we werestarting to talk about earlier
(44:34):
their name gets bought and soldin ways that they can't control
and in ways that feel really badfor them.
Then the follow on effectsespecially if you're a college
that is trading on this brand oflike we give a shit about you
and we want to get to know you,but you've met them through this
mechanism that feels really bad, like that's a broken system.
(44:56):
So if you take that same smallexample about how students get
connected to colleges and youthink about how to do that
better, you let students controlit.
You let students say yes or noto being connected with a
particular college and becausethey control it, they are much
more willing to share with uscandidly about who they are,
(45:18):
what motivates them, what theirpreferences are, and those
preferences and data points notonly help us understand what
their challenges are and whatproducts we should create for
them, but helps us advise ourcollege partners about what
their population of studentsreally care about.
So there is a genuineness andcandor component that actually
(45:39):
makes those product decisionsreally easy, because then you
get to just drive behaviors outof those sort of that felt
belief or that held belief.
The behaviors and tactics thatyou deploy just feel really
simple, all in service of that.
So I don't know that it'sactually that complicated with
that at the center, because thatis the thing that frees us up
(46:03):
to really then focus on like,okay, what do we innovate on?
Where can we gain efficiencies?
What might we makerecommendations to a college to
actually do?
Speaker 1 (46:11):
Yeah, I love that answer and just letting your I
think values and beliefs reallysort of guide some of your
decision-making there.
You've been super generous withyour I think values and beliefs
really sort of guide some ofyour decision-making there.
You've been super generous withyour time.
Like, what are you excitedabout?
Like there's so much sort ofhappening in this space.
You all are building somepretty incredible things.
Like what gets you personallyexcited?
Speaker 2 (46:32):
I am personally excited for like solving more
stuff for students.
I can remember being a collegestudent and not knowing how to
run, you know, my college searchjourney Once I got to college
being really confused when I hadto go to the registrar and ask
for a transcript, notunderstanding the systems that I
(46:52):
was a part of, and in my day itwas like, you know, going to
talk to like old Marge in theregistrar's office and you had
to go do something in triplicateand then you had to like mail
it here and put it in this slotand like it just seemed like a
lot of sort of like magic that Ididn't really understand how I
was part of it.
And I think that problem stillexists today, where students
(47:14):
actually have a higherexpectation of how they're
treated like as consumers.
Right To like that we shouldknow more about them and treat
them better.
And I think the stakes arehigher for institutions right,
because with more and moreinstitutions closing and being
under like serious financialduress and that's bad for
communities, we have to sort ofdo two things One is like become
(47:36):
more efficient as an industryand also like meet a higher
standard and I actually thinkthat better use of AI with data
underneath it is the key leverto how that's going to happen,
and I can see this in, like youknow, if I think about, if I put
my sort of like student searchhat on as an enrollment marketer
(47:56):
, I'm pretty happy with a schoolthat buys a list of names and
then you market to that list.
You get like a 19% open rate andI think that's pretty good.
When I think about what AI isdoing, right now, it's closer to
a 40% open rate Right, so Itrack that over time and then I
watch the efficiencies thatstart to fall out of that, which
(48:17):
is like, oh, we could run asmaller funnel.
We could actually do this moreefficiently, find better fit
students and create classesabout the things that we want.
Like, another layer of what Ican do on sort of the class
building sides of things is helpyou index on like we are a
college that creates leaders andwe need a really civic minded
(48:41):
class.
That's a job that you can haveAI go do across so much data
that we like cannot hold in ourhuman brains.
Speaker 1 (48:50):
Oh yeah.
Speaker 2 (48:50):
So I think overall, like the solving of that higher
standard that students have andmaking it more efficient at the
same time and sort of that likebig orchestration of all of the
things that need to happen tomake those two things true, are
what I'm just like super excitedto be a part of, because in my
journey I got to see from Margein the registrar's office and
(49:12):
the frigging triplicate piecesof paper to like what we can do
today and it's just thrilling towatch that continue.
Speaker 1 (49:21):
Couldn't agree more.
I think that's a fabulous placeto end it.
But, yeah, thank you for thetime here, emily, and yeah,
please let people know how theycan get in touch with you.
Speaker 2 (49:32):
Yeah, so you can find me on collegevinecom, on
LinkedIn, at Emily Smith, andyou can also check out.
I also have a podcast, which isreally similar to this one.
So if you like the topics thatMike talks about, I too have a
podcast.
It's called the Weekly VineDown.
You can find us atthevinedowncom and I hang out
(49:53):
there putting out a new showabout every week.
Speaker 1 (49:56):
Awesome.
Well, this has been awesome.
Thanks again.
Speaker 2 (49:59):
Thanks for having me, Mike.
Popular Podcasts
Bookmarked by Reese's Book Club
Welcome to Bookmarked by Reese’s Book Club — the podcast where great stories, bold women, and irresistible conversations collide! Hosted by award-winning journalist Danielle Robay, each week new episodes balance thoughtful literary insight with the fervor of buzzy book trends, pop culture and more. Bookmarked brings together celebrities, tastemakers, influencers and authors from Reese's Book Club and beyond to share stories that transcend the page. Pull up a chair. You’re not just listening — you’re part of the conversation.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com