July 29, 2025 • 44 mins
The Advisory Board Podcast — Featuring Tushar Mishra, Co-Founder & CEO of Delightree
Episode sponsored by ClientTether — big thanks to them for supporting the franchise community.
In this episode, host Dave Hansen sits down with Tushar Mishra, Co-Founder and CEO of Delightree, to explore how franchisors can stop talking about AI and actually start using it. From automating franchisee support to enabling data-driven coaching, Tushar shares practical, no-hype strategies to make AI an everyday operational advantage—not just a buzzword.
You’ll also hear how Delightree’s tech-first approach is streamlining multi-unit operations across 5,000+ locations, and why your operations team might be sitting on a goldmine of untapped efficiencies.
Here’s what you’ll learn:
AI Without the Fluff: Tushar breaks down what’s actually working in franchise systems today—from SOP-powered AI search to real-time support deflection.
Stop Answering the Same Question Twice: Learn how franchisors are cutting 50% of incoming franchisee questions by transforming ops manuals and training docs into smart, searchable systems.
Just-in-Time Learning: Gen Z doesn’t want e-learning PDFs. Tushar explains how Delightree is enabling short, actionable, multilingual training that fits the modern workforce.
FBCs Supercharged: Discover how AI can prep franchise business coaches with instant, data-backed insights—so they coach smarter, not harder.
Data-Driven Onboarding: With mapped rule-based workflows, AI companions, and actionable triggers, Delightree’s onboarding toolkit ensures new owners ramp up with clarity and confidence.
Predictive Coaching & Churn Prevention: Tushar unpacks how AI tools can flag underperforming units before they go off the rails—and how you can course-correct early using signals from POS data, audits, reviews, and training compliance.
Memorable Quotes:
“Half of support tickets come from franchisees asking for information they already have—but can’t find. AI solves that instantly.”
“AI doesn’t replace coaching. It just makes your best coach 10x more effective.”
“Don’t start with AI. Start with a business problem. The right AI will follow.”
Connect with Tushar:
📧 tushar@delightree.com
🔗 delightree.com
Don’t miss this one if you're ready to ditch the AI hype and finally make it useful—especially for multi-unit ops, support-heavy systems, and FBCs running on caffeine and spreadsheets. Tushar’s insights are smart, scalable, and ready to implement today.
#franchiseops #aiforfranchising #franchisegrowth #delightree #clienttether
Episode sponsored by ClientTether — big thanks to them for supporting the franchise community.
In this episode, host Dave Hansen sits down with Tushar Mishra, Co-Founder and CEO of Delightree, to explore how franchisors can stop talking about AI and actually start using it. From automating franchisee support to enabling data-driven coaching, Tushar shares practical, no-hype strategies to make AI an everyday operational advantage—not just a buzzword.
You’ll also hear how Delightree’s tech-first approach is streamlining multi-unit operations across 5,000+ locations, and why your operations team might be sitting on a goldmine of untapped efficiencies.
Here’s what you’ll learn:
AI Without the Fluff: Tushar breaks down what’s actually working in franchise systems today—from SOP-powered AI search to real-time support deflection.
Stop Answering the Same Question Twice: Learn how franchisors are cutting 50% of incoming franchisee questions by transforming ops manuals and training docs into smart, searchable systems.
Just-in-Time Learning: Gen Z doesn’t want e-learning PDFs. Tushar explains how Delightree is enabling short, actionable, multilingual training that fits the modern workforce.
FBCs Supercharged: Discover how AI can prep franchise business coaches with instant, data-backed insights—so they coach smarter, not harder.
Data-Driven Onboarding: With mapped rule-based workflows, AI companions, and actionable triggers, Delightree’s onboarding toolkit ensures new owners ramp up with clarity and confidence.
Predictive Coaching & Churn Prevention: Tushar unpacks how AI tools can flag underperforming units before they go off the rails—and how you can course-correct early using signals from POS data, audits, reviews, and training compliance.
Memorable Quotes:
“Half of support tickets come from franchisees asking for information they already have—but can’t find. AI solves that instantly.”
“AI doesn’t replace coaching. It just makes your best coach 10x more effective.”
“Don’t start with AI. Start with a business problem. The right AI will follow.”
Connect with Tushar:
📧 tushar@delightree.com
🔗 delightree.com
Don’t miss this one if you're ready to ditch the AI hype and finally make it useful—especially for multi-unit ops, support-heavy systems, and FBCs running on caffeine and spreadsheets. Tushar’s insights are smart, scalable, and ready to implement today.
#franchiseops #aiforfranchising #franchisegrowth #delightree #clienttether
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Guys, I want to welcome you to another episode of
the Franchise Advisory Board, where we bring experts in from
around the franchise industry, suppliers, brands and anybody else anybody
between to give you actionable, practical advice that can help
you to grow your franchise brand. And I've got with
me a friend of mine, Tashar Metrom and he is
the co founder and CEO of a brand called delight Tree.
(00:23):
If you're not familiar with them, they are the fastest
growing franchisee operations platform in the industry. Incredible mobile first.
We could nerd out about this maybe in another episode
right about what they do, but they're interesting because they
were built very in a very good framework that enables
franchises to dip right into the platform and get benefit
(00:43):
immediately from day one. Think where a client oether goes
from like franchise lead to close deal, anything that happens
after you on board or you sign a franchise agreement.
Everything needs to happen after that, store openers, things like that.
That's what they've built here, including lms's and all sorts
of great great tools. Was got Ai infused into it.
Now a couple of other things you may not know
(01:06):
about to shore and which he was reluctant to tell me,
but believe it or not, he was actually a heavy
metal and still as a heavy metal enthusiast used to
have long hair thrashed and headbanged. Like if you if
you ever hang out with him, you would be like,
no freaking way, because he's so calm and brit and
just relaxed. But that to Sharre, I'm grateful that you're here.
Thanks for giving us something. Tell us more about you,
(01:26):
tell us about delight Tree, and then we'll jump into
our topic today.
Speaker 2 (01:30):
Oh absolutely, thanks, thanks a lot, Dave for having me.
My name is the Sharp, one of the co founders
of delight Tree. Actually, the way I ended up in
franchising is pretty interesting. Before delight Tree, I used to
run a company called Servador, and that me and my
co founders to my co founder here at delight Tree,
started right out of college and that was like our
project Viva Nerds. We did like a lot of NLP
(01:52):
data analytics and stuff, and we built out this algorithm
that would analyze on structure texts data and we built
a bunch of skapers and my last company We used
to help restaurant chains, hotel chains, retail chains and aggregating
their customer reviews from different channels and giving them one
dashboard where they could sort of look at those reviews
and get some analytics around them. A while we were
(02:13):
sort of running that company, working with about three thousand businesses,
what we noticed was that even though we were giving
out these insights that we thought were extremely helpful, none
of our customers were actually taking actions on it. And
the reason was what when we investigated, and we learned,
in most companies, you have the marketing teams. We're looking
at these dashboard, is there something interesting? They'll pass that
(02:35):
out to the op steam of steam, to the regional manager,
digional manager's store owners, storeround with guy working there, and
there's no tracking of this happening. So that sort of
got us super intrigued. And then when we exited Surveytor
instead of that led to the genesis of Delight three
and have been in love with franchising and multi unit
businesses ever since.
Speaker 1 (02:55):
Yeah, well, you can describe what happens every day in franchising.
Great insight comes from the data and analytics gets passed
down to owners, gets passed down to the frontline workers
and then sometimes nothing happens. So yeah, you found a
good sweet spot where you can apply your skills to
help fix that problem.
Speaker 2 (03:11):
Absolutely, no, it's been fun. So as we start building
the lightly, we knew that the first thing that we
had to do was just talk to as many franchisors
and as many franchises as we can.
Speaker 3 (03:22):
And so the first.
Speaker 2 (03:23):
Product, I mean a lot of people don't notice, but
the first product that we built was just task management
for franchising owners. And so a lot of our customers
were just somebody who owned a few subway locations, seven
eleven locations and they could basically sit at their couch
and run like an opening and closing checklists and see
if everything happened on time. And from that, when we
launched that in twenty twenty one, to sort of just
(03:46):
getting a meted in the industry tending I face, going
to as many events as we want, talking to amazing
set of leaders like you, we eventually learned instead of
build the platform that we have today, which is again
the fastest growing platform within Chasing Helping now ninety plus brans,
so close to five thousand locations. Now that we work
with though pretty exciting and a lot of exciting things
(04:09):
coming so before to have the industry so.
Speaker 1 (04:12):
Yeah, well I've seen you guys grow very quickly, so
it's been fun to see. And man, let's dive into
some of the topic today with everyone. To Shar, we
want to talk about the AI driven franchise and this
is guys, I'll just warn you this is not a
hype discussion. This is actual pragmatic stuff that's really working
to Shar season from the back end of dozens and
(04:33):
dozens of brands, So I hope you'll enjoy the insights
he's got to share. Really, what we want to talk
about is how do you use AI to scale franchise
operations and profits. A lot of people, I mean us included,
we're talking about how do you use AI to improve
sales processes, et cetera. But really it feels like kind
of the bastard child of the industry sometimes with the
attention and the investment and the money and the head count.
(04:55):
Is this operation side that really has a lot of
opportunity to leverage AI. So to Shark, could you maybe
start off by introducing us a little bit to this
concept of how what are some practical strategies for the
way people could be leveraging AI on the franchise operations
to create more success.
Speaker 3 (05:13):
Now, I think and you gave a great intro.
Speaker 2 (05:16):
Like AI has been an important hype topic in the
industry for a while. Every event that I go to
there are at least a few speakers are talking about AI.
When it comes to practice, the only actual use case
that I've seen is people using chat GPT to create
marketing collaterals, or on the consumer end where people are
(05:39):
maybe using automated callers and detecting calls and stuff like that. Right,
And what's unfortunate about this is that there's so much
that EI can do today that directly helps your ops
profitability and margins at the sore level that most people
either don't know about, don't speak about, or just don't
know how to implement.
Speaker 3 (05:59):
Right.
Speaker 2 (05:59):
So I thought this could be an interesting topic where
you're not just discussing sort of what's going to happen
in the future with things that brands can do today.
And this is not talking about you have to use
lightly do them, but today with tools that already exist. Right.
And so with that being said, I want to start
with the first thing, right. So when it comes to
(06:20):
AI or anything AI, the most important thing that you
as a business would have is data. So the first
place to start here is within your OPS. If you
don't have data, which means that you don't have SOPs
built out, you don't have any measurement of how franchises
are performing from an OPS perspective. So that could be
(06:41):
things like how frequent are the completing tasks, their franchise
audits or store visits, scorecards, your Google ratings, your employee attrition.
Maybe at this Z level, all of this is going
to be hard, right, So I think the first step
is just getting a sense of what all data that
(07:02):
you have and where all does it exist? Right, So
that becomes sort of the fundamental layer over which you'll
sort of do everything. Now from there, let's let's let's
really start to break down operations from a things to
do or things that you're supposed to.
Speaker 3 (07:18):
Do perspective, right.
Speaker 2 (07:19):
So the first thing that comes to mind that a
lot of time OPS teams are investing on is answering questions.
And we've seen a bunch of solutions pop up that
help you sort of automate support in the sense and
in this case. I'm specifically talking about your franchise owner's
(07:42):
owner asks you, Hey, this ad campaign that you're running
this month, where do I find the collacter?
Speaker 1 (07:48):
Is it on my email? Is it on my drive?
Where is it?
Speaker 2 (07:52):
And you call somebody, you get in touch with them,
and stead of you get an output.
Speaker 3 (07:56):
So the first use that I.
Speaker 2 (07:58):
Know some other companies in the set are also doing
Easier system is a great example that likely does it
a bunch of other companies to it is really creating
a data warehouse or a data lake of all of
your SOPs so that you can automate answering some of
these questions.
Speaker 3 (08:14):
Right.
Speaker 2 (08:15):
What that would mean is that when your owners are
looking for an answer, your first layer of support becomes
the answer that is generated from the material that already exists.
And this is something that's already in practice majority of
our customers, so we within Delight, we have a tool
called AI Search. Majority of our customers use AI Search
(08:36):
and they have been able to deflect over half of
support requests and queries that come from owners directly to HQO. Right,
all the little things like how to do this, how
to do that? Where do I find this? All of
that can be automated.
Speaker 1 (08:52):
Wow, So I just want to go back for a
second and make sure that everyone understands what you're just
what you're describing. Guys, you know your OPS manuals is
everything you can feel, feed it into the back end
of an AI engine, and then make sure that you
and feed a bunch of FAQs so understands how you
want it to answer, like tone, style, et cetera, and
then it can answer these questions for you. But I
(09:13):
just want to highlight what you just said. You just
said that for your clients using this tool set, this
AAI is a way to answer questions for franchises, you're
reducing the amount of questions actually need to go to
the franchise or by fifty percent.
Speaker 2 (09:28):
Absolutely, any FPC that you talked to, they would say
that half of the request or half of the questions
that they get are things that already exist, but people
don't know where to find them, or sometimes even SBCs
don't know where to find them. Right, And so you
can automate like a bulk of that and these are
all the questions about where how that already existed the system.
(09:51):
And the best part about doing this is any request
that you're getting that does not exist in the system,
yet you can now create a dot for that, and
over time you're not just reducing it by fifty percent,
but you're reducing it by sixty percent, eighty percent. What
time will just be five percent of those requests coming
to you. One way that I like to think about
(10:12):
this and our team sort of talks to a lot
of franchisors every day and we try to coach them,
especially on these support tickets. Right, So if you look
at support that owners typically want, they fall in three
main categories. So one is when owners are asking just
about where certain things.
Speaker 3 (10:31):
Is this right?
Speaker 2 (10:31):
They just want to find where's this training, where's this collateral?
Where do I find this specific information? The second category
typically is about how I do certain things right? So
that's more about how do I make sure that I've
hired like a new bunch of people, how do I
make sure that they don't chure or I'm running this campaign,
(10:55):
how do I optimize it right? Things like that. And
then third category is more request basis right, so which
means that when owners are asking you to do something.
So I have a golf simulator franchisee, and I have
a request that my simulator is not working and I
need somebody from EAHQU to come and help me out.
(11:15):
The first category can today one hundred percent be automated
to the right. Over time you can add more information
and that gets automated. The second category, which is a
little more nuanced that requires more data, can again fifty
percent or so be automated.
Speaker 1 (11:31):
Today.
Speaker 2 (11:32):
We have a new and I want to maybe talk
about that towards the end, but we have a new
module that's coming up that also automates a lot of
insight generation and coaching specifically for owners. You can do that,
and so the only thing now that you have to
take care of is requests that are coming for actions
that you have to write. So if you think of
it from that framework, and think for each of these
(11:53):
buckets in my framework, what data exists and what all
do I need to create in order to start automation
and do that? And maybe one more thing that I
want to add here. A lot of people when they're
thinking about this, they think about tooling, right, what do
I use in order to do that? Obviously there are
some franchise these specific solutions exist to like the easiest
(12:13):
is that a bunch of others, But if you don't
want to use that, if you still want to start
and experiment this, my recommendation is make a custom GPT,
add some of your sort of request on that, and
make that custom GPT available to some of your team
members and start to experiment with it. Once you start
to see that the support request going down and people
(12:36):
are getting value from it, then start to look out
for a vendor.
Speaker 1 (12:40):
That's a great because and go ahead, go ahead to sure.
Speaker 2 (12:46):
I was just saying a lot of times that that
specific sort of technique might not scale because you need permissions.
And for example, you don't want to give that access
for that engine to somebody who's in the frontline team,
or say somebody who does not need that information, Like
within my corporate team. I don't want to give like
a marketing intern access to this so that they can
(13:08):
search up, Hey, what were how do I collect my
royalties and what we're sort of quorities for last month
and stuff like that. So when you want to scale,
you obviously want more permissioning and you want sort of
tighter security and stuff, and so you might want to
go for a vendor. But before that, just to experiment,
I mean, you can do it on your own leg
in a day, yeah.
Speaker 1 (13:26):
Yeah, in an hour. If you know what you're doing
and either a perplexity, CHT, GPT whatever, you can build
that out fast. Yeah, love it well. And that's it's
kind of like, well anyway in medical world, Hey you
got a back problems, put a court zone shot in there,
and if that makes it feel better, you know that
we need to fix something. And that's that's what you're doing.
You're just kind of you're very very low brow testing
(13:49):
the concept and proving that it's going to have an impact.
I love that and absolutely I want to highlight here, guys,
from what I've seen work well in the industry, this
works franchisees across all aspects of their business. If it doesn't,
and it doesn't matter what kind of business. If you're
a home service business, if you're a restaurant, if you're
any kind of bit a franchise model, a gym, a
(14:11):
nutrition store, it doesn't matter. Franchisees have to have basic
questions answered. And this is just an operating necessity and
if you want to be able to you know that.
I know the end the industry, they say you need
probably one f FBC per thirty brands. When you're younger,
it's a little bit lower ratio because there are a
lot more kinks you're working out. But I want to
go back to something to Shar said. Here you can
(14:33):
upload your OPS manual and whatever FAQ's you've got, and
that's a good starting point, but that's only going to
get you to maybe fifty percent uh to Shar mentioned this,
but this is just key to AI keeping it, not
just keeping it fresh, but keeping it keep continuing to
close the gap between what is the AI know and
how can it help? And what is what is what
is your team know? How do you recommend to Shar
(14:54):
that they build a process around adding new content? You
mentioned this briefly, but I want to go back. This
is a key step in really leveraging AI and building
out what we used to call knowledge bases back in
the day, which were self served kind of wikis. This
is a way to really streamline that whole flow. But
there's a methodology and a discipline to that. Describe that
to the listeners today so they know how do they
(15:16):
continue to close the gap using a tool like this
for ai Q and.
Speaker 2 (15:19):
A no, that's a that's a great question, right, and
every time we talk to a new business, and I'm
specially talking about business a certain scale. Obviously when you
when you're small, you're still trying to figure figure stuff out.
But even at larger businesses, you have this data extremely siloed.
So you might have here, and I'll give you an example.
A lot of people will listen to this and say,
(15:40):
oh should we We also have like a similar set
of models. So you might have your OPS manual in
one place, which gives details around everything how you open,
how you close, how you do certain things. But think
about different tools, products, machines that you use in their manuals. Right, So,
for example, if it's a restaurant, you have a point
(16:02):
of sale system, owners have questions about how to do
certain things. Hey, this paper is not printing, how do
I comp a customer, etc?
Speaker 3 (16:08):
Where does that live?
Speaker 2 (16:10):
Right? The third category then becomes, Okay, I have my
OPS manual, I have sort of manuals of everything else
in my store. This could be a machine, this could
be a software system. But in addition to that, do
I also have local laws compliance regulation that might be
more industry specific. If it's a restaurant, do I also
(16:30):
have knowledge based off asset and OSHA and different things
that develop into certain franchises and connect cater them to
their location. So for some location in California, California is
more element rather than sort of federal OSHA right, some
of these things. So my recommendation is first just take
(16:50):
inventory of what you have, just and be honest with
yourself for what you have. And that's why I started
by saying even large scale brands don't have set of
this data figured out. But it's not that hard to
sort of prepare this. It'll probably take you a few
days to get inventory of all of them. Once you
have them, create a one repository of those and that
(17:12):
could be in today's world. Lllms are so good at
like ingesting information that could be any format. So you
could have some documents in PDF, thumb in words and
could just be images, whatever you have, just put them
in one single place. That could be a drive that
could be whatever you wanted. My next step is, especially
when you want to scale, if you're doing an experiment,
probably that is enough. You feed that into a custom
(17:34):
GPT and try that out with a few trusted people
or franchises that you have and play around with it.
Once you want to scale. One important thing that you
have to do in addition to that is label that data.
Right now, what that means is, I'm not talking about
fancy data labelaying off what this means, what that means,
but just label who this is relevant to and who
(17:54):
can access this?
Speaker 3 (17:56):
Right.
Speaker 2 (17:56):
So if I have my manual, every including frontline teams,
should be able to access that versus certain parts in
my manual, just the owners should be able to access right.
And so doing this is really important and would help
you because you are basically teaching an LM. Think of
LM as a child. You're teaching an element LLM. Who
(18:18):
can you answer what to and who should you not
answer when they ask a specific question?
Speaker 3 (18:24):
Right?
Speaker 2 (18:25):
And we've heard these stories when companies got over excited
and start to use sort of AI internally and somebody
very junior could ask, hey, give me a list of
all my customers, right if you give them access to
POS and you teach them how to do that. So
you want to avoid some of those mistakes. So I
think on data, do that first, and then second is labeling.
(18:45):
If you're working with a vendor, they can really help
you do that and really defined. For example, on the lightree.
We have a knowledge base will load school that way
where you can dump all the information in any format
that you have. Could be videos, could be audio, could
be anything.
Speaker 3 (18:58):
That you have.
Speaker 2 (18:58):
And we just ingested make the LLLM learn according to
whoever is asking the question so that they can get
revelpment responses and whoever the vendor that you're working with
public can help you do that.
Speaker 1 (19:10):
Yeah, that's amazing, But most people don't think about labeling
the data as far as access and also relevance. So
how do you teach this this ll M that that
it's going to be producing a relevant result? That really
you got to You got to teach it and then
after time it it'll start to learn on its own.
That's that's the key. Stuff to train it. That's great.
(19:30):
You know one of the things that you shared to
shar that I really like is is thinking down the road,
like how can you know AI is great? How do
you make it better? Really? That's that's the key is AI.
The more specialized it is and the more you train it,
the better and better and better it can get. And
that's not I mean that's been true for fifteen years
with with with AI back before we used to call
(19:51):
it that. So tell me a little bit about about
this so when we're when we're trying to so we've
talked about KNADA, the f a q q A and
unsallowing the data getting even like your manuals in there.
Even one thing I want to share on that is
before before we go on, I forgot this is a
key thought is you have different rules of operation in
different states. Uh, And that's tricky, especially things like let's
(20:15):
just talk about like searcharging right, search charging is illegal
in a bunch of states, and in certain states you
have to follow certain protocols. If you're going to do surchargings,
you have to offer different payment methods. In Georgia or else,
it's illegal to surcharge anything. And there's a limit in
some states of how much they can so feeding into
your into the database, into the end of the AI
module that you're choosing to use, like the geographic information.
(20:38):
What where the laws change, Like you've got to be specific,
And especially as you talked about California, it made me
think of this, like so many rules are so different
in California. You've just upload a general operations manual, it's
going to give bad advice to franchise owners because it
won't know what's relevant for a geography. So you also
have to make sure because we're operating different states with
different rules, that you've you've got, you've got that data
(21:00):
are tagged inside the system so the system can spit
back accurate data or it knows it has to ask
a qualifying question or which state do you operate in
then it can give you relevant data. So those are
things I just to add to to sharsad to make
sure you're getting.
Speaker 2 (21:14):
Absolutely Yeah, that's a that's a great point. Then I
think before you close the thought, I think the last
thing to really hone down on, especially on the Q
and A answering and all that bit is think about
what is the form factor or where do your employees.
Speaker 3 (21:32):
Or your owners access this?
Speaker 2 (21:35):
Do you want them? If you're experimenting that's just a
custom GPT, you can create a team account, add your
seat of a few owners that you're testing it with,
and just use sort of on the mobile or on
the web app chat GPT to sort of do that.
But if you want to do it at scale, you
also have to think about am I introducing another tool
now that people would sort of query. If they get
(21:59):
a response to do X, Now where does the action
for X happen? Is that another tool that I'm using?
So think of where do people access it? Are you
solving the problem or are you creating one more problem
now where people have to figure out that of okay,
I ask you, but then I do there? Or it
(22:20):
says okay, it existed this drive and I go and
try to access it, but I don't have permission to
sort of access the drive. And so think about some
of these things as well, especially on the accessibility side.
How do you actually make it usable for your team?
Speaker 1 (22:35):
Yeah, very relevant. Point in fact, it's just just as
important as managing who's got access to what data is?
What's the experience? Where? How do they how do they
gain access to it? And I'll give you an example.
It comes to my mind if you if you want
to well, let's let's say a franchise owner is trying
to access is this custom GPT we're talking about. This
(22:57):
is where scalability comes in a great They go to
the special link that you provide them, They say, can
open up a chat session and get an answer. While
in that answer, it turns out that the GPT can
answer the question, Well, now they have to log into
another system to submit a ticket to this whole process.
If happened to exit one system go to another system,
(23:18):
it creates a lag but also a drop off point.
Sometimes people don't have a whole lot of patients for that,
and so they'll just say screw it, I'll figure it
out of my own or I'll do it later, and
then they forget so they never end up asking the question.
It's critical for their business, So that form factor of saying,
how do we converge? How do we create a uniform
experience where I can get answers to my questions through
(23:39):
through this AI tool and if I can't, then I
can immediately submit a ticket, or I can immediately get
in contact with the team member that I need to
or like. That type of an experience is critical to
make sure AI fits in alongside the whole support infrastructure.
You're not creating yet another silo of technology. Yep, yep.
Absolutely to shre there a bunch of things that we
(24:00):
ought to see if we can talk about today, but
I'm guessing we're probably gonna run out of time, So
let's what if you don't mind, maybe let's talk about
two other areas where you've seen AI that can be
used very well. One is in automating franchisee onboarding and training.
And we've talked about the kind of the q and
as the daily minutia of the fbcs, but let's talk
about training and onboarding, and let's talk about using AI
(24:21):
to predict franchise e performance and even reducing churn. You
kind of hinted that a few minutes ago, but I
think we had to talk about that. So if you
mind sharing, like, what are you seeing what's working in
each of those areas, let's start first with the franchisee
onboarding and training.
Speaker 2 (24:36):
That's a yeah, no, I mean both of these spaces
are so interesting and so a few people are on end.
I mean it gets me very excited. We can noddu
on this a lot. But let's start with onboarding and training, right.
And So when it comes to franchise onboarding or opening
a new location, most businesses have a very viewed process, right.
So they will have something internally, maybe a project management
(24:58):
solution that they're using, but there's nothing that the owner
or the new owner is interfacing with it. Imagine that's
the first experience that the owner has coming to the brand.
You want to make sure that's consistence, that's tied. They
can get any help that they need, But today's sort
of that's broken, right. And then second sort of onboarding
in a second part of onboarding is training. The one
(25:22):
you first need to figure out how to train the owners,
but you also need to think about how do I
empower this owner to now train employees and not just
once but probably every quarter or every month because the
attrition is going to be high depending on the industry
that you're in, right, And so there are a few
things that we've seen people get a lot of success,
(25:43):
especially on the onboarding and training. But let's talk about
onboarding first, creating a rule set which clearly maps out
your onboarding process. For example, this needs to happen then
that then that this specific thing cannot happen without this.
For example, I cannot get start on my training before
I get my fire.
Speaker 3 (26:05):
Safety certification, right, and that could be one of your processes.
Speaker 2 (26:09):
Now, how do you sort of map that out and
roll that out to your franchise becomes important and where
AI can help is that now instead of the new
prospective owner being clueless about what to do, next, if
they have a companion that they can talk to and
can tell them exactly what to do next, why they
cannot do certain things, and why certain things have to
(26:31):
be done before doing a certain thing, right, And then
you can go one step further.
Speaker 3 (26:36):
And that do add more details on each process.
Speaker 2 (26:39):
Let's take this fire safety sort of check example, right,
and so I'm an owner and I need to get
a fire safety inspection done before I can attend the
straining that the brand is built out. Now the agent
would tell me, hey, you need to compete this first,
I don't know how.
Speaker 1 (26:55):
To do it.
Speaker 2 (26:56):
Here are two preferred vendors that the brand listed. You
can to them to sort of get the inspection done
or get the right resources. And then once that's done,
it automatically triggers somebody in my team to sort of
kick off my training. So that's when you can create
a really tight sort of onboarding experience for new owners.
It also scales, right, so that doesn't mean that you
(27:18):
need new people. And for every new franchise that you add,
you probably need to add resources it needs. With one
resource properly mapped out, you can add hundreds and hundreds
of franchises that are opening sort of on time, on schedule,
and with proper experience for the owners and your team members. Second,
specifically on training, one big disconnect sort of in the
(27:42):
industry about training is E learning became popular, right, and
everybody sort of jumped on that train and created these
beautiful looking courses that you have to go through in
order to learn that became available to frontline teams. The
problem with that is that's not how most people learn today, right.
If you talk to a eighteen year old, they would say, Hey,
(28:05):
the way that I learn is I skim through information
and then when I need to, I'll just search Google
or ask JRGBT or to search that information right and
get it when I need to. And that's something what
we call just in time training. Now, with AI, you
can really customize just in time training for your new
owners in a way that they can learn not just faster,
(28:26):
but better, which ultimately delivers a better sort of experience
to your end user or your end customer. As an example,
I start off working sort of as a cashier. I
need to learn everything about the point of sale system. Now,
instead of going through that large E learning course and
try to remember everything, what if a I can just
(28:47):
recommend nine of the times you'll be doing five of
these things. I'll teach you these five things so that
you know, okay, how to add order, how to cancel order,
how to like here's the that you need to do.
So I'm going to learn that everything else. When I
need to, I can just ask and quickly learn. So
now there's a special case somebody screwed up and I
(29:09):
need to comp the customer and I don't want them
to charge anything.
Speaker 3 (29:13):
I don't know how to do that.
Speaker 2 (29:15):
Pures where I can use it to quickly ask, Right,
if there's sort of a device in the store. Some
of our customers are experimenting with that. So if you
use a tool like lightly, you can just ask HEDI,
how do I comp a customer and get a response quickly,
and you do that. So you're not spending a lot
of time and just paying employees to learn when you
know they're not actually learning but getting them to do work,
(29:36):
and then as they're working, they'll learn as they go.
Another interesting thing that we do specifically on lightly is
learning with actions, right, so you're not just watching a
bunch of videos and answering questions, but as you are
sort of on your first day, on your second day,
on your third day, you're doing actions, clicking pictures, adding
videos that you're doing by learning, and then your colleague
(29:58):
instead of mark them, evaluate them.
Speaker 3 (30:00):
So if you go in become sort of employee.
Speaker 1 (30:05):
I love that. So I don't know if you know this,
but I spend much time in the L and D industry,
and then I spend a much time in the language
industry localizing and globalizing people's learning content. And I just
want to back up what Tishar is saying, like it's
it's important that you make your learning modules interactive. So
many of the times like here, read this pdf. Guess
what a eighteen year old's never going to do in
their life read a pdf to learn how to do something.
Speaker 2 (30:27):
They just want to do it.
Speaker 1 (30:28):
And you have to. So you have to serve up
microscopic data parcels right like you have to. You have
to be able to break it down in a way
that they can consume it. And even a three minute
video segment is a turnoff, Like you got to figure
out what is the learning by, how would I make
this bite size? What are the learning components? And then
you have to have a module something they'll recommend I
love that using the AI to say, look, ninety percent
(30:50):
of the people they need these five things. So I'm
going to recommend you watch those first. If that doesn't
answer your question, ask me something else. Like I love
the idea of infusing AI into the back end of
the data to learn how people are engaging with the
content and serving up the most relevant content. First. Guess
what does that already? And what this next generation is
used to Apple music, every single thing they consume, right, Netflix?
Speaker 3 (31:13):
Everything, netflixs YouTube?
Speaker 1 (31:14):
Yeah. Yeah, So it's learn to be the same, you know,
learning to be.
Speaker 2 (31:20):
Absolutely And one thing I'll add here is especially on
like when you mentioned languages and localization. I think that's
very important. Today that problem is solved. You can have
your learning courses, teaching courses in any language that you want,
and I can access them in any language that I'm
comfortable in. So I could be at a store and
customers do that with delight freet today. You could be
(31:41):
at a store you don't know how to speak English,
you just speak Spanish. All of my SOPs, my manual,
my material are stritten in English.
Speaker 3 (31:50):
But I can still.
Speaker 2 (31:51):
Ask a question in Spanish and get a response in
Spanish and still get my work done. And that is
I mean, you don't need to use lightly for that.
There's so many tools sort of to help you do
that today. But that is a solve problem today. So
you can actually have multiple languages, courses, people speaking them,
and you can just get stuff done.
Speaker 1 (32:09):
Yeah, it's amazing. It didn't exist before, didn't exist before
tell us. In the last few minutes, we've got to
sharpe tell us a little bit about how can people
use like how should people start thinking about using AI
to predict franchise e performance and even churn?
Speaker 2 (32:25):
That is that is my favorite topic and maybe let's
let's spend some time on it. So the first thing
that you have to do in order to actually get
this right and actually start to use this is going
to my first point, get data in place.
Speaker 3 (32:38):
It's going to be.
Speaker 2 (32:39):
Hard if you have one tool that you're using for
doing audits, another tool that you're using for tasks, and
other tools that you're using as your elems, and then
you have something else for scheduling. You need to figure
out how to either get them to talk to each
other or create a data warehouse where you can dump
all the data coming in from all these tools. So
(33:00):
let's just assume that you have this data warehouse where
you've dumped all the data that you have. And when
I say all the data, I mean any tools and
system that you use. It includes your POS that includes
quick books that your franchise owners are using, where your
P and L lies everything.
Speaker 3 (33:16):
Now, once you have that, and some.
Speaker 2 (33:18):
People have already might have tried that, have you ever
tried uploading like a Excel file on CHATGPT and asking
you to analyze that and giving you some insights out
of that. It's actually pretty good at that now. So
four point one, which is specifically an analytics model, can
actually look at insights and tell you exactly what's hidden
(33:40):
in the data. Now you can actually do that for
franchises at scale. Traditionally, fbcs have been sort of focused
in doing that. So if you're a good FPC or
if you're an excellent FBC, what you would do is
before meeting each franchiseing, you'll make sure that you know
everything about them. You have the right data at the
right stuff. So when you're coaching, you're actually coaching and
(34:00):
not just hey you did this wrong or you need
to do this better. Then turn that into a blame
game today. What you can do is so imagine if
I'm in FBC, before I go to a meeting, I
just ask AI, hey, what should I be talking about today?
What a I would do is it search through my
(34:20):
last audit score. It searts through my revenue compared that
to the same time last year. Compare that to my peers.
You look at my attrition, you look at my Google reviews,
you look at task completion, and look at training completion,
all of that and find patterns and tell me exactly
what I need to talk about. So my response could
(34:40):
be that this specific location scored very low on these
specific things on an audit. Say they scored low on
cleanliness on an audit. By the way, Google Reviews are
also talking about that, and within their peers their revenue
is lower than all of them. Right now, I have
proper talking points and data that I can take to
(35:03):
the fpcup to the owner and talk to them about
with actual action deble a sort of insights that they
can do today. So, okay, cleanness is an issue because
I see attrition is really high with your staff members.
Why is attrition high? Oh, I see training completion rate
is low? Okay, what kind of training are you doing. Oh,
by the way, this training, we have this new training
that you can introduce to sort of do that now
(35:25):
instead of just saying that my revenue is down or
my store is not clean and getting to the root
cause of that problem and actually coaching the owner and
working with the owner to solve the problem.
Speaker 3 (35:35):
Right. That's just one example.
Speaker 2 (35:37):
Now, this is something that a lot of people have
been wanted to do, a lot of enterprise are doing.
Seems like way into the future, but it's already, it's
already there. We have a module right now that we
are piloting with a few customers that we launch soon
that does exactly this, right, which is how do you
amend your FPC. So, going back to your first point shirt,
(36:00):
for every thirty vacations have one FBC, I say, why
not for every three hundred location you have one FBC.
But that one FBC now can do that job really, really,
really well. Right. And So, and this is again to
one of the flaws for when people get excited about AI,
they want to automate everything, and I've never seen it
(36:21):
working when you try to. Again, you don't want to
automate the coaching layer. You don't want that personal touch
to go away. Somebody who's a partner to you, especially
an owner, you don't want them to talk to a
robot all the time. You want an FBC. You want
a person, but you want to empower that person now
to be able to do that job ten x faster
or ten x easier, so that they can do it
(36:42):
with ten x more locations.
Speaker 3 (36:44):
Right.
Speaker 2 (36:46):
Yeah, I know, I'll pause here. We can talk about
this all day, but this bit is pretty exciting, and
I see in the near future, I think in the
next couple of years, more and more franchises who's trought
to use this and you see this tectonic and how
the relationship between Z and so it plays out because
it'll be less about he said, she said, and more
(37:07):
about sort of actual data that you could both work
towards solving.
Speaker 1 (37:11):
It's funny, you said, I was going to interject and
say something similar in that we we live in a
world right now where franchise ors and franchisees often have
a lot of tension between them, and sometimes because we're
not speaking from the same playbook, the franchise sees it
from one perspective. Franchise or sees it from another. But
the one thing that is universal and is not arguable
(37:32):
is the data. Right. So if you're using objection objective
data to give coaching advice or or a behavior change
or behavior modification training, whatever you're trying to do with it,
it's a lot harder for the franchise to push back
on it. And we see that from our perspective on
the selling side, the franchise franchise e performance site on
revenue generation all the time, the FBC's are relieved that
(37:55):
they can look at data and then call and say, hey,
it looks like compared to your peers, blah blah blah. Right,
but being able to amalgamate that data and then have
the AI essentially spit out almost a script for the
FBC to say, look, you need to call Bob about X,
y Z, like these are the three things Bob needs
help with, And this how Bob stacks up against his
peers in region. This how Bob stacks up against his
(38:16):
peers and his age demographic, like how long they've all
been franchise owners? Like for I know for some franchise
systems there are sometimes one or two full time people
trying to generate that data and information so they can
have insights to coach with. This can all be done
very easily with a button click. Now, if like Tashar said,
you get your data lined up so then AI can
(38:38):
read it and interpret it and take action on it,
and then give your fbcs what they need to take
their actions, and don't just a personal touch like you.
We can't lose that because people, as much as AI
is awesome, people still want people to interact with. You
don't get any sort of emotional kick from the or
chemical kick in your brain from AI interactions like you do.
Speaker 2 (39:00):
People absolutely absolutely, and I know I'm repeating myself, but
I'll say these two things again. So you have to
be careful about who can access that data. So you
need that sort of labeling on the data side, and
really well, especially when you have this much data, and
it could be dangerous right like when people can access
everything or you're letting AI access everything, So how do
(39:22):
you get that properly to certain people, to certain roles,
to certain context. And second, when you have the insight, now,
how do you connect that with actions? Think about both
these things.
Speaker 3 (39:33):
If you're just.
Speaker 2 (39:34):
Getting that split out response in chat GPT, are you
able to is that changing anything for you, You're still
going to do the same thing. I'm going to send
a slab to my OPS manager, my OPPS manager is
going to send a text message to my digional manager,
regional manager to the owner and there it's not going anywhere.
So also think of how do you close the loop
(39:55):
by actually bringing actions into the fold and what are
some tools and what's that can help you do that
in the more it's consistent way creating a unified solution.
Speaker 1 (40:05):
It's great feedback. Yeah, because at the end of the day,
all is data. All these insights, they are meaningless if
they don't change behavior, and that's the hard part. So
beautiful and that's where the human personal touch can come
in as well as Hey, here is the insight. This
is what we talked about, Steve, I don't see you've
taken any action on that, Like, how do you know?
You can see it in the data? So I love that.
(40:27):
And that's where coaches can really coach, right, their whole
title is to coach, But you're not doing a lot
of coaching if all you're doing is data analysis and
chasing your tail answering basic questions, which again AI can
do way better than them anyway.
Speaker 3 (40:40):
Yep, absolutely, Well, I.
Speaker 1 (40:43):
Know, Tashar, we're a little tight on time. Now, tell
me any any parting thoughts on where you anticipate people
can see the greatest lift right now from leveraging AI
in their franchise operations. Like if they had to pick
one of the three things we talked about today, where
should somebody start?
Speaker 2 (41:00):
So that's a great question, and I'll start by saying
where not to start? Right? So where not to start
is you're getting excited about AI and everybody is using AI,
and I have this pomo and I just want to
use AI. Right, don't do that unless you're starting with
a proper business outcome, and we have done that. Everybody
has done that. We have to start with the business outcome.
(41:22):
Pick out whatever is the first priority for you. Some
of these are hard to do, some of these are
easier to do. Some of these united you need the
right vendor. Obviously, I want to say, reach out to
us and we can help you do all of this,
but that's that's not the point of this. I would say,
pick out what one, which one is priority one for you,
and then also rank them in order of difficulty, which
(41:44):
means I can do the answering instead of that bit.
But I don't have data, so it's going to be
very difficult for me. I can do the training bit
or it's not a priority for me, but it's easier
to do, right, So do that, create a matrix, see
what is HRN priority and easier to do, and start
with that. I always recommend start experimenting in a very
(42:05):
low cost manner. Today you can do it absolutely for free.
Experiment with a few of your owners and see that's
actually helping you get to the business outcomes. For example,
if I want to automate my support, I can again
create a customer GPT thealthree of my owners. Hey, we're
experimenting with this. Let's try using this for a week
and see if like your questions are getting answered or not.
(42:27):
They're getting answered, and I can seat of e and
put it in a sheet that a number of requests
I was getting before number of requests and getting Now
now you know it's solving the problem and you can
allocate budget to it, which makes sense. So you can say, okay,
I want to scale this to all one hundred locations
and I can now properly create that sort of budget
and what it would mean for me if I can
(42:50):
reach that outcome. So that's that's what I would say,
just to start with the business outcome, not just with AI,
and go step by step manner that a lot of
flex smart people I know DA if they can reach
out to you, if they need help, they can reach
out to me. But yeah, with some of this stuff,
it's easy to get started, like at your home, just
(43:10):
on your computer.
Speaker 1 (43:13):
I agree. That's one of the beauties of what's shifting
in AI. It's kind of like access to being literate.
Back in the day, Like, yeah, nobody had access to
learn how to read, and so only very few people
had access to the data. But it was like that
with AI, only specific people at a massive amount of
money could access AI. And now it's not that way.
Anyone can do it. Yes, I'd love that to Shart.
(43:34):
You've given us a lot of good insights today. If
somebody wants to reach out to you or connect with you,
what's the best way for them to do that.
Speaker 2 (43:41):
The best way is go to delight free dot com,
d E L I G H t R E E
dot com. Reach out to us. We will be available
if you want to reach out directly to me. It's
my first name. To shore p U s h a
r At delight free dot com. I answered almost to
all my emails, So you have any questions, you don't
(44:01):
have to talk about the lightly. Even if you want
to talk about AI and how you set up a
certain experiment, I'm happy to help.
Speaker 1 (44:06):
Yeah, I figured you'd say something like that, Dashar, You're awesome.
Thanks for making time to share some of those insights
with us and get a little bit deeper into reality
within AI and how it can be applied. I sure
appreciate you doing that today.
Speaker 3 (44:19):
Absolutely thanks. Thanks about you. Was be chatting with you
Guys, I want to welcome you to another episode of
the Franchise Advisory Board, where we bring experts in from
around the franchise industry, suppliers, brands and anybody else anybody
between to give you actionable, practical advice that can help
you to grow your franchise brand. And I've got with
me a friend of mine, Tashar Metrom and he is
the co founder and CEO of a brand called delight Tree.
(00:23):
If you're not familiar with them, they are the fastest
growing franchisee operations platform in the industry. Incredible mobile first.
We could nerd out about this maybe in another episode
right about what they do, but they're interesting because they
were built very in a very good framework that enables
franchises to dip right into the platform and get benefit
(00:43):
immediately from day one. Think where a client oether goes
from like franchise lead to close deal, anything that happens
after you on board or you sign a franchise agreement.
Everything needs to happen after that, store openers, things like that.
That's what they've built here, including lms's and all sorts
of great great tools. Was got Ai infused into it.
Now a couple of other things you may not know
(01:06):
about to shore and which he was reluctant to tell me,
but believe it or not, he was actually a heavy
metal and still as a heavy metal enthusiast used to
have long hair thrashed and headbanged. Like if you if
you ever hang out with him, you would be like,
no freaking way, because he's so calm and brit and
just relaxed. But that to Sharre, I'm grateful that you're here.
Thanks for giving us something. Tell us more about you,
(01:26):
tell us about delight Tree, and then we'll jump into
our topic today.
Speaker 2 (01:30):
Oh absolutely, thanks, thanks a lot, Dave for having me.
My name is the Sharp, one of the co founders
of delight Tree. Actually, the way I ended up in
franchising is pretty interesting. Before delight Tree, I used to
run a company called Servador, and that me and my
co founders to my co founder here at delight Tree,
started right out of college and that was like our
project Viva Nerds. We did like a lot of NLP
(01:52):
data analytics and stuff, and we built out this algorithm
that would analyze on structure texts data and we built
a bunch of skapers and my last company We used
to help restaurant chains, hotel chains, retail chains and aggregating
their customer reviews from different channels and giving them one
dashboard where they could sort of look at those reviews
and get some analytics around them. A while we were
(02:13):
sort of running that company, working with about three thousand businesses,
what we noticed was that even though we were giving
out these insights that we thought were extremely helpful, none
of our customers were actually taking actions on it. And
the reason was what when we investigated, and we learned,
in most companies, you have the marketing teams. We're looking
at these dashboard, is there something interesting? They'll pass that
(02:35):
out to the op steam of steam, to the regional manager,
digional manager's store owners, storeround with guy working there, and
there's no tracking of this happening. So that sort of
got us super intrigued. And then when we exited Surveytor
instead of that led to the genesis of Delight three
and have been in love with franchising and multi unit
businesses ever since.
Speaker 1 (02:55):
Yeah, well, you can describe what happens every day in franchising.
Great insight comes from the data and analytics gets passed
down to owners, gets passed down to the frontline workers
and then sometimes nothing happens. So yeah, you found a
good sweet spot where you can apply your skills to
help fix that problem.
Speaker 2 (03:11):
Absolutely, no, it's been fun. So as we start building
the lightly, we knew that the first thing that we
had to do was just talk to as many franchisors
and as many franchises as we can.
Speaker 3 (03:22):
And so the first.
Speaker 2 (03:23):
Product, I mean a lot of people don't notice, but
the first product that we built was just task management
for franchising owners. And so a lot of our customers
were just somebody who owned a few subway locations, seven
eleven locations and they could basically sit at their couch
and run like an opening and closing checklists and see
if everything happened on time. And from that, when we
launched that in twenty twenty one, to sort of just
(03:46):
getting a meted in the industry tending I face, going
to as many events as we want, talking to amazing
set of leaders like you, we eventually learned instead of
build the platform that we have today, which is again
the fastest growing platform within Chasing Helping now ninety plus brans,
so close to five thousand locations. Now that we work
with though pretty exciting and a lot of exciting things
(04:09):
coming so before to have the industry so.
Speaker 1 (04:12):
Yeah, well I've seen you guys grow very quickly, so
it's been fun to see. And man, let's dive into
some of the topic today with everyone. To Shar, we
want to talk about the AI driven franchise and this
is guys, I'll just warn you this is not a
hype discussion. This is actual pragmatic stuff that's really working
to Shar season from the back end of dozens and
(04:33):
dozens of brands, So I hope you'll enjoy the insights
he's got to share. Really, what we want to talk
about is how do you use AI to scale franchise
operations and profits. A lot of people, I mean us included,
we're talking about how do you use AI to improve
sales processes, et cetera. But really it feels like kind
of the bastard child of the industry sometimes with the
attention and the investment and the money and the head count.
(04:55):
Is this operation side that really has a lot of
opportunity to leverage AI. So to Shark, could you maybe
start off by introducing us a little bit to this
concept of how what are some practical strategies for the
way people could be leveraging AI on the franchise operations
to create more success.
Speaker 3 (05:13):
Now, I think and you gave a great intro.
Speaker 2 (05:16):
Like AI has been an important hype topic in the
industry for a while. Every event that I go to
there are at least a few speakers are talking about AI.
When it comes to practice, the only actual use case
that I've seen is people using chat GPT to create
marketing collaterals, or on the consumer end where people are
(05:39):
maybe using automated callers and detecting calls and stuff like that. Right,
And what's unfortunate about this is that there's so much
that EI can do today that directly helps your ops
profitability and margins at the sore level that most people
either don't know about, don't speak about, or just don't
know how to implement.
Speaker 3 (05:59):
Right.
Speaker 2 (05:59):
So I thought this could be an interesting topic where
you're not just discussing sort of what's going to happen
in the future with things that brands can do today.
And this is not talking about you have to use
lightly do them, but today with tools that already exist. Right.
And so with that being said, I want to start
with the first thing, right. So when it comes to
(06:20):
AI or anything AI, the most important thing that you
as a business would have is data. So the first
place to start here is within your OPS. If you
don't have data, which means that you don't have SOPs
built out, you don't have any measurement of how franchises
are performing from an OPS perspective. So that could be
(06:41):
things like how frequent are the completing tasks, their franchise
audits or store visits, scorecards, your Google ratings, your employee attrition.
Maybe at this Z level, all of this is going
to be hard, right, So I think the first step
is just getting a sense of what all data that
(07:02):
you have and where all does it exist? Right, So
that becomes sort of the fundamental layer over which you'll
sort of do everything. Now from there, let's let's let's
really start to break down operations from a things to
do or things that you're supposed to.
Speaker 3 (07:18):
Do perspective, right.
Speaker 2 (07:19):
So the first thing that comes to mind that a
lot of time OPS teams are investing on is answering questions.
And we've seen a bunch of solutions pop up that
help you sort of automate support in the sense and
in this case. I'm specifically talking about your franchise owner's
(07:42):
owner asks you, Hey, this ad campaign that you're running
this month, where do I find the collacter?
Speaker 1 (07:48):
Is it on my email? Is it on my drive?
Where is it?
Speaker 2 (07:52):
And you call somebody, you get in touch with them,
and stead of you get an output.
Speaker 3 (07:56):
So the first use that I.
Speaker 2 (07:58):
Know some other companies in the set are also doing
Easier system is a great example that likely does it
a bunch of other companies to it is really creating
a data warehouse or a data lake of all of
your SOPs so that you can automate answering some of
these questions.
Speaker 3 (08:14):
Right.
Speaker 2 (08:15):
What that would mean is that when your owners are
looking for an answer, your first layer of support becomes
the answer that is generated from the material that already exists.
And this is something that's already in practice majority of
our customers, so we within Delight, we have a tool
called AI Search. Majority of our customers use AI Search
(08:36):
and they have been able to deflect over half of
support requests and queries that come from owners directly to HQO. Right,
all the little things like how to do this, how
to do that? Where do I find this? All of
that can be automated.
Speaker 1 (08:52):
Wow, So I just want to go back for a
second and make sure that everyone understands what you're just
what you're describing. Guys, you know your OPS manuals is
everything you can feel, feed it into the back end
of an AI engine, and then make sure that you
and feed a bunch of FAQs so understands how you
want it to answer, like tone, style, et cetera, and
then it can answer these questions for you. But I
(09:13):
just want to highlight what you just said. You just
said that for your clients using this tool set, this
AAI is a way to answer questions for franchises, you're
reducing the amount of questions actually need to go to
the franchise or by fifty percent.
Speaker 2 (09:28):
Absolutely, any FPC that you talked to, they would say
that half of the request or half of the questions
that they get are things that already exist, but people
don't know where to find them, or sometimes even SBCs
don't know where to find them. Right, And so you
can automate like a bulk of that and these are
all the questions about where how that already existed the system.
(09:51):
And the best part about doing this is any request
that you're getting that does not exist in the system,
yet you can now create a dot for that, and
over time you're not just reducing it by fifty percent,
but you're reducing it by sixty percent, eighty percent. What
time will just be five percent of those requests coming
to you. One way that I like to think about
(10:12):
this and our team sort of talks to a lot
of franchisors every day and we try to coach them,
especially on these support tickets. Right, So if you look
at support that owners typically want, they fall in three
main categories. So one is when owners are asking just
about where certain things.
Speaker 3 (10:31):
Is this right?
Speaker 2 (10:31):
They just want to find where's this training, where's this collateral?
Where do I find this specific information? The second category
typically is about how I do certain things right? So
that's more about how do I make sure that I've
hired like a new bunch of people, how do I
make sure that they don't chure or I'm running this campaign,
(10:55):
how do I optimize it right? Things like that. And
then third category is more request basis right, so which
means that when owners are asking you to do something.
So I have a golf simulator franchisee, and I have
a request that my simulator is not working and I
need somebody from EAHQU to come and help me out.
(11:15):
The first category can today one hundred percent be automated
to the right. Over time you can add more information
and that gets automated. The second category, which is a
little more nuanced that requires more data, can again fifty
percent or so be automated.
Speaker 1 (11:31):
Today.
Speaker 2 (11:32):
We have a new and I want to maybe talk
about that towards the end, but we have a new
module that's coming up that also automates a lot of
insight generation and coaching specifically for owners. You can do that,
and so the only thing now that you have to
take care of is requests that are coming for actions
that you have to write. So if you think of
it from that framework, and think for each of these
(11:53):
buckets in my framework, what data exists and what all
do I need to create in order to start automation
and do that? And maybe one more thing that I
want to add here. A lot of people when they're
thinking about this, they think about tooling, right, what do
I use in order to do that? Obviously there are
some franchise these specific solutions exist to like the easiest
(12:13):
is that a bunch of others, But if you don't
want to use that, if you still want to start
and experiment this, my recommendation is make a custom GPT,
add some of your sort of request on that, and
make that custom GPT available to some of your team
members and start to experiment with it. Once you start
to see that the support request going down and people
(12:36):
are getting value from it, then start to look out
for a vendor.
Speaker 1 (12:40):
That's a great because and go ahead, go ahead to sure.
Speaker 2 (12:46):
I was just saying a lot of times that that
specific sort of technique might not scale because you need permissions.
And for example, you don't want to give that access
for that engine to somebody who's in the frontline team,
or say somebody who does not need that information, Like
within my corporate team. I don't want to give like
a marketing intern access to this so that they can
(13:08):
search up, Hey, what were how do I collect my
royalties and what we're sort of quorities for last month
and stuff like that. So when you want to scale,
you obviously want more permissioning and you want sort of
tighter security and stuff, and so you might want to
go for a vendor. But before that, just to experiment,
I mean, you can do it on your own leg
in a day, yeah.
Speaker 1 (13:26):
Yeah, in an hour. If you know what you're doing
and either a perplexity, CHT, GPT whatever, you can build
that out fast. Yeah, love it well. And that's it's
kind of like, well anyway in medical world, Hey you
got a back problems, put a court zone shot in there,
and if that makes it feel better, you know that
we need to fix something. And that's that's what you're doing.
You're just kind of you're very very low brow testing
(13:49):
the concept and proving that it's going to have an impact.
I love that and absolutely I want to highlight here, guys,
from what I've seen work well in the industry, this
works franchisees across all aspects of their business. If it doesn't,
and it doesn't matter what kind of business. If you're
a home service business, if you're a restaurant, if you're
any kind of bit a franchise model, a gym, a
(14:11):
nutrition store, it doesn't matter. Franchisees have to have basic
questions answered. And this is just an operating necessity and
if you want to be able to you know that.
I know the end the industry, they say you need
probably one f FBC per thirty brands. When you're younger,
it's a little bit lower ratio because there are a
lot more kinks you're working out. But I want to
go back to something to Shar said. Here you can
(14:33):
upload your OPS manual and whatever FAQ's you've got, and
that's a good starting point, but that's only going to
get you to maybe fifty percent uh to Shar mentioned this,
but this is just key to AI keeping it, not
just keeping it fresh, but keeping it keep continuing to
close the gap between what is the AI know and
how can it help? And what is what is what
is your team know? How do you recommend to Shar
(14:54):
that they build a process around adding new content? You
mentioned this briefly, but I want to go back. This
is a key step in really leveraging AI and building
out what we used to call knowledge bases back in
the day, which were self served kind of wikis. This
is a way to really streamline that whole flow. But
there's a methodology and a discipline to that. Describe that
to the listeners today so they know how do they
(15:16):
continue to close the gap using a tool like this
for ai Q and.
Speaker 2 (15:19):
A no, that's a that's a great question, right, and
every time we talk to a new business, and I'm
specially talking about business a certain scale. Obviously when you
when you're small, you're still trying to figure figure stuff out.
But even at larger businesses, you have this data extremely siloed.
So you might have here, and I'll give you an example.
A lot of people will listen to this and say,
(15:40):
oh should we We also have like a similar set
of models. So you might have your OPS manual in
one place, which gives details around everything how you open,
how you close, how you do certain things. But think
about different tools, products, machines that you use in their manuals. Right, So,
for example, if it's a restaurant, you have a point
(16:02):
of sale system, owners have questions about how to do
certain things. Hey, this paper is not printing, how do
I comp a customer, etc?
Speaker 3 (16:08):
Where does that live?
Speaker 2 (16:10):
Right? The third category then becomes, Okay, I have my
OPS manual, I have sort of manuals of everything else
in my store. This could be a machine, this could
be a software system. But in addition to that, do
I also have local laws compliance regulation that might be
more industry specific. If it's a restaurant, do I also
(16:30):
have knowledge based off asset and OSHA and different things
that develop into certain franchises and connect cater them to
their location. So for some location in California, California is
more element rather than sort of federal OSHA right, some
of these things. So my recommendation is first just take
(16:50):
inventory of what you have, just and be honest with
yourself for what you have. And that's why I started
by saying even large scale brands don't have set of
this data figured out. But it's not that hard to
sort of prepare this. It'll probably take you a few
days to get inventory of all of them. Once you
have them, create a one repository of those and that
(17:12):
could be in today's world. Lllms are so good at
like ingesting information that could be any format. So you
could have some documents in PDF, thumb in words and
could just be images, whatever you have, just put them
in one single place. That could be a drive that
could be whatever you wanted. My next step is, especially
when you want to scale, if you're doing an experiment,
probably that is enough. You feed that into a custom
(17:34):
GPT and try that out with a few trusted people
or franchises that you have and play around with it.
Once you want to scale. One important thing that you
have to do in addition to that is label that data.
Right now, what that means is, I'm not talking about
fancy data labelaying off what this means, what that means,
but just label who this is relevant to and who
(17:54):
can access this?
Speaker 3 (17:56):
Right.
Speaker 2 (17:56):
So if I have my manual, every including frontline teams,
should be able to access that versus certain parts in
my manual, just the owners should be able to access right.
And so doing this is really important and would help
you because you are basically teaching an LM. Think of
LM as a child. You're teaching an element LLM. Who
(18:18):
can you answer what to and who should you not
answer when they ask a specific question?
Speaker 3 (18:24):
Right?
Speaker 2 (18:25):
And we've heard these stories when companies got over excited
and start to use sort of AI internally and somebody
very junior could ask, hey, give me a list of
all my customers, right if you give them access to
POS and you teach them how to do that. So
you want to avoid some of those mistakes. So I
think on data, do that first, and then second is labeling.
(18:45):
If you're working with a vendor, they can really help
you do that and really defined. For example, on the lightree.
We have a knowledge base will load school that way
where you can dump all the information in any format
that you have. Could be videos, could be audio, could
be anything.
Speaker 3 (18:58):
That you have.
Speaker 2 (18:58):
And we just ingested make the LLLM learn according to
whoever is asking the question so that they can get
revelpment responses and whoever the vendor that you're working with
public can help you do that.
Speaker 1 (19:10):
Yeah, that's amazing, But most people don't think about labeling
the data as far as access and also relevance. So
how do you teach this this ll M that that
it's going to be producing a relevant result? That really
you got to You got to teach it and then
after time it it'll start to learn on its own.
That's that's the key. Stuff to train it. That's great.
(19:30):
You know one of the things that you shared to
shar that I really like is is thinking down the road,
like how can you know AI is great? How do
you make it better? Really? That's that's the key is AI.
The more specialized it is and the more you train it,
the better and better and better it can get. And
that's not I mean that's been true for fifteen years
with with with AI back before we used to call
(19:51):
it that. So tell me a little bit about about
this so when we're when we're trying to so we've
talked about KNADA, the f a q q A and
unsallowing the data getting even like your manuals in there.
Even one thing I want to share on that is
before before we go on, I forgot this is a
key thought is you have different rules of operation in
different states. Uh, And that's tricky, especially things like let's
(20:15):
just talk about like searcharging right, search charging is illegal
in a bunch of states, and in certain states you
have to follow certain protocols. If you're going to do surchargings,
you have to offer different payment methods. In Georgia or else,
it's illegal to surcharge anything. And there's a limit in
some states of how much they can so feeding into
your into the database, into the end of the AI
module that you're choosing to use, like the geographic information.
(20:38):
What where the laws change, Like you've got to be specific,
And especially as you talked about California, it made me
think of this, like so many rules are so different
in California. You've just upload a general operations manual, it's
going to give bad advice to franchise owners because it
won't know what's relevant for a geography. So you also
have to make sure because we're operating different states with
different rules, that you've you've got, you've got that data
(21:00):
are tagged inside the system so the system can spit
back accurate data or it knows it has to ask
a qualifying question or which state do you operate in
then it can give you relevant data. So those are
things I just to add to to sharsad to make
sure you're getting.
Speaker 2 (21:14):
Absolutely Yeah, that's a that's a great point. Then I
think before you close the thought, I think the last
thing to really hone down on, especially on the Q
and A answering and all that bit is think about
what is the form factor or where do your employees.
Speaker 3 (21:32):
Or your owners access this?
Speaker 2 (21:35):
Do you want them? If you're experimenting that's just a
custom GPT, you can create a team account, add your
seat of a few owners that you're testing it with,
and just use sort of on the mobile or on
the web app chat GPT to sort of do that.
But if you want to do it at scale, you
also have to think about am I introducing another tool
now that people would sort of query. If they get
(21:59):
a response to do X, Now where does the action
for X happen? Is that another tool that I'm using?
So think of where do people access it? Are you
solving the problem or are you creating one more problem
now where people have to figure out that of okay,
I ask you, but then I do there? Or it
(22:20):
says okay, it existed this drive and I go and
try to access it, but I don't have permission to
sort of access the drive. And so think about some
of these things as well, especially on the accessibility side.
How do you actually make it usable for your team?
Speaker 1 (22:35):
Yeah, very relevant. Point in fact, it's just just as
important as managing who's got access to what data is?
What's the experience? Where? How do they how do they
gain access to it? And I'll give you an example.
It comes to my mind if you if you want
to well, let's let's say a franchise owner is trying
to access is this custom GPT we're talking about. This
(22:57):
is where scalability comes in a great They go to
the special link that you provide them, They say, can
open up a chat session and get an answer. While
in that answer, it turns out that the GPT can
answer the question, Well, now they have to log into
another system to submit a ticket to this whole process.
If happened to exit one system go to another system,
(23:18):
it creates a lag but also a drop off point.
Sometimes people don't have a whole lot of patients for that,
and so they'll just say screw it, I'll figure it
out of my own or I'll do it later, and
then they forget so they never end up asking the question.
It's critical for their business, So that form factor of saying,
how do we converge? How do we create a uniform
experience where I can get answers to my questions through
(23:39):
through this AI tool and if I can't, then I
can immediately submit a ticket, or I can immediately get
in contact with the team member that I need to
or like. That type of an experience is critical to
make sure AI fits in alongside the whole support infrastructure.
You're not creating yet another silo of technology. Yep, yep.
Absolutely to shre there a bunch of things that we
(24:00):
ought to see if we can talk about today, but
I'm guessing we're probably gonna run out of time, So
let's what if you don't mind, maybe let's talk about
two other areas where you've seen AI that can be
used very well. One is in automating franchisee onboarding and training.
And we've talked about the kind of the q and
as the daily minutia of the fbcs, but let's talk
about training and onboarding, and let's talk about using AI
(24:21):
to predict franchise e performance and even reducing churn. You
kind of hinted that a few minutes ago, but I
think we had to talk about that. So if you
mind sharing, like, what are you seeing what's working in
each of those areas, let's start first with the franchisee
onboarding and training.
Speaker 2 (24:36):
That's a yeah, no, I mean both of these spaces
are so interesting and so a few people are on end.
I mean it gets me very excited. We can noddu
on this a lot. But let's start with onboarding and training, right.
And So when it comes to franchise onboarding or opening
a new location, most businesses have a very viewed process, right.
So they will have something internally, maybe a project management
(24:58):
solution that they're using, but there's nothing that the owner
or the new owner is interfacing with it. Imagine that's
the first experience that the owner has coming to the brand.
You want to make sure that's consistence, that's tied. They
can get any help that they need, But today's sort
of that's broken, right. And then second sort of onboarding
in a second part of onboarding is training. The one
(25:22):
you first need to figure out how to train the owners,
but you also need to think about how do I
empower this owner to now train employees and not just
once but probably every quarter or every month because the
attrition is going to be high depending on the industry
that you're in, right, And so there are a few
things that we've seen people get a lot of success,
(25:43):
especially on the onboarding and training. But let's talk about
onboarding first, creating a rule set which clearly maps out
your onboarding process. For example, this needs to happen then
that then that this specific thing cannot happen without this.
For example, I cannot get start on my training before
I get my fire.
Speaker 3 (26:05):
Safety certification, right, and that could be one of your processes.
Speaker 2 (26:09):
Now, how do you sort of map that out and
roll that out to your franchise becomes important and where
AI can help is that now instead of the new
prospective owner being clueless about what to do, next, if
they have a companion that they can talk to and
can tell them exactly what to do next, why they
cannot do certain things, and why certain things have to
(26:31):
be done before doing a certain thing, right, And then
you can go one step further.
Speaker 3 (26:36):
And that do add more details on each process.
Speaker 2 (26:39):
Let's take this fire safety sort of check example, right,
and so I'm an owner and I need to get
a fire safety inspection done before I can attend the
straining that the brand is built out. Now the agent
would tell me, hey, you need to compete this first,
I don't know how.
Speaker 1 (26:55):
To do it.
Speaker 2 (26:56):
Here are two preferred vendors that the brand listed. You
can to them to sort of get the inspection done
or get the right resources. And then once that's done,
it automatically triggers somebody in my team to sort of
kick off my training. So that's when you can create
a really tight sort of onboarding experience for new owners.
It also scales, right, so that doesn't mean that you
(27:18):
need new people. And for every new franchise that you add,
you probably need to add resources it needs. With one
resource properly mapped out, you can add hundreds and hundreds
of franchises that are opening sort of on time, on schedule,
and with proper experience for the owners and your team members. Second,
specifically on training, one big disconnect sort of in the
(27:42):
industry about training is E learning became popular, right, and
everybody sort of jumped on that train and created these
beautiful looking courses that you have to go through in
order to learn that became available to frontline teams. The
problem with that is that's not how most people learn today, right.
If you talk to a eighteen year old, they would say, Hey,
(28:05):
the way that I learn is I skim through information
and then when I need to, I'll just search Google
or ask JRGBT or to search that information right and
get it when I need to. And that's something what
we call just in time training. Now, with AI, you
can really customize just in time training for your new
owners in a way that they can learn not just faster,
(28:26):
but better, which ultimately delivers a better sort of experience
to your end user or your end customer. As an example,
I start off working sort of as a cashier. I
need to learn everything about the point of sale system. Now,
instead of going through that large E learning course and
try to remember everything, what if a I can just
(28:47):
recommend nine of the times you'll be doing five of
these things. I'll teach you these five things so that
you know, okay, how to add order, how to cancel order,
how to like here's the that you need to do.
So I'm going to learn that everything else. When I
need to, I can just ask and quickly learn. So
now there's a special case somebody screwed up and I
(29:09):
need to comp the customer and I don't want them
to charge anything.
Speaker 3 (29:13):
I don't know how to do that.
Speaker 2 (29:15):
Pures where I can use it to quickly ask, Right,
if there's sort of a device in the store. Some
of our customers are experimenting with that. So if you
use a tool like lightly, you can just ask HEDI,
how do I comp a customer and get a response quickly,
and you do that. So you're not spending a lot
of time and just paying employees to learn when you
know they're not actually learning but getting them to do work,
(29:36):
and then as they're working, they'll learn as they go.
Another interesting thing that we do specifically on lightly is
learning with actions, right, so you're not just watching a
bunch of videos and answering questions, but as you are
sort of on your first day, on your second day,
on your third day, you're doing actions, clicking pictures, adding
videos that you're doing by learning, and then your colleague
(29:58):
instead of mark them, evaluate them.
Speaker 3 (30:00):
So if you go in become sort of employee.
Speaker 1 (30:05):
I love that. So I don't know if you know this,
but I spend much time in the L and D industry,
and then I spend a much time in the language
industry localizing and globalizing people's learning content. And I just
want to back up what Tishar is saying, like it's
it's important that you make your learning modules interactive. So
many of the times like here, read this pdf. Guess
what a eighteen year old's never going to do in
their life read a pdf to learn how to do something.
Speaker 2 (30:27):
They just want to do it.
Speaker 1 (30:28):
And you have to. So you have to serve up
microscopic data parcels right like you have to. You have
to be able to break it down in a way
that they can consume it. And even a three minute
video segment is a turnoff, Like you got to figure
out what is the learning by, how would I make
this bite size? What are the learning components? And then
you have to have a module something they'll recommend I
love that using the AI to say, look, ninety percent
(30:50):
of the people they need these five things. So I'm
going to recommend you watch those first. If that doesn't
answer your question, ask me something else. Like I love
the idea of infusing AI into the back end of
the data to learn how people are engaging with the
content and serving up the most relevant content. First. Guess
what does that already? And what this next generation is
used to Apple music, every single thing they consume, right, Netflix?
Speaker 3 (31:13):
Everything, netflixs YouTube?
Speaker 1 (31:14):
Yeah. Yeah, So it's learn to be the same, you know,
learning to be.
Speaker 2 (31:20):
Absolutely And one thing I'll add here is especially on
like when you mentioned languages and localization. I think that's
very important. Today that problem is solved. You can have
your learning courses, teaching courses in any language that you want,
and I can access them in any language that I'm
comfortable in. So I could be at a store and
customers do that with delight freet today. You could be
(31:41):
at a store you don't know how to speak English,
you just speak Spanish. All of my SOPs, my manual,
my material are stritten in English.
Speaker 3 (31:50):
But I can still.
Speaker 2 (31:51):
Ask a question in Spanish and get a response in
Spanish and still get my work done. And that is
I mean, you don't need to use lightly for that.
There's so many tools sort of to help you do
that today. But that is a solve problem today. So
you can actually have multiple languages, courses, people speaking them,
and you can just get stuff done.
Speaker 1 (32:09):
Yeah, it's amazing. It didn't exist before, didn't exist before
tell us. In the last few minutes, we've got to
sharpe tell us a little bit about how can people
use like how should people start thinking about using AI
to predict franchise e performance and even churn?
Speaker 2 (32:25):
That is that is my favorite topic and maybe let's
let's spend some time on it. So the first thing
that you have to do in order to actually get
this right and actually start to use this is going
to my first point, get data in place.
Speaker 3 (32:38):
It's going to be.
Speaker 2 (32:39):
Hard if you have one tool that you're using for
doing audits, another tool that you're using for tasks, and
other tools that you're using as your elems, and then
you have something else for scheduling. You need to figure
out how to either get them to talk to each
other or create a data warehouse where you can dump
all the data coming in from all these tools. So
(33:00):
let's just assume that you have this data warehouse where
you've dumped all the data that you have. And when
I say all the data, I mean any tools and
system that you use. It includes your POS that includes
quick books that your franchise owners are using, where your
P and L lies everything.
Speaker 3 (33:16):
Now, once you have that, and some.
Speaker 2 (33:18):
People have already might have tried that, have you ever
tried uploading like a Excel file on CHATGPT and asking
you to analyze that and giving you some insights out
of that. It's actually pretty good at that now. So
four point one, which is specifically an analytics model, can
actually look at insights and tell you exactly what's hidden
(33:40):
in the data. Now you can actually do that for
franchises at scale. Traditionally, fbcs have been sort of focused
in doing that. So if you're a good FPC or
if you're an excellent FBC, what you would do is
before meeting each franchiseing, you'll make sure that you know
everything about them. You have the right data at the
right stuff. So when you're coaching, you're actually coaching and
(34:00):
not just hey you did this wrong or you need
to do this better. Then turn that into a blame
game today. What you can do is so imagine if
I'm in FBC, before I go to a meeting, I
just ask AI, hey, what should I be talking about today?
What a I would do is it search through my
(34:20):
last audit score. It searts through my revenue compared that
to the same time last year. Compare that to my peers.
You look at my attrition, you look at my Google reviews,
you look at task completion, and look at training completion,
all of that and find patterns and tell me exactly
what I need to talk about. So my response could
(34:40):
be that this specific location scored very low on these
specific things on an audit. Say they scored low on
cleanliness on an audit. By the way, Google Reviews are
also talking about that, and within their peers their revenue
is lower than all of them. Right now, I have
proper talking points and data that I can take to
(35:03):
the fpcup to the owner and talk to them about
with actual action deble a sort of insights that they
can do today. So, okay, cleanness is an issue because
I see attrition is really high with your staff members.
Why is attrition high? Oh, I see training completion rate
is low? Okay, what kind of training are you doing. Oh,
by the way, this training, we have this new training
that you can introduce to sort of do that now
(35:25):
instead of just saying that my revenue is down or
my store is not clean and getting to the root
cause of that problem and actually coaching the owner and
working with the owner to solve the problem.
Speaker 3 (35:35):
Right. That's just one example.
Speaker 2 (35:37):
Now, this is something that a lot of people have
been wanted to do, a lot of enterprise are doing.
Seems like way into the future, but it's already, it's
already there. We have a module right now that we
are piloting with a few customers that we launch soon
that does exactly this, right, which is how do you
amend your FPC. So, going back to your first point shirt,
(36:00):
for every thirty vacations have one FBC, I say, why
not for every three hundred location you have one FBC.
But that one FBC now can do that job really, really,
really well. Right. And So, and this is again to
one of the flaws for when people get excited about AI,
they want to automate everything, and I've never seen it
(36:21):
working when you try to. Again, you don't want to
automate the coaching layer. You don't want that personal touch
to go away. Somebody who's a partner to you, especially
an owner, you don't want them to talk to a
robot all the time. You want an FBC. You want
a person, but you want to empower that person now
to be able to do that job ten x faster
or ten x easier, so that they can do it
(36:42):
with ten x more locations.
Speaker 3 (36:44):
Right.
Speaker 2 (36:46):
Yeah, I know, I'll pause here. We can talk about
this all day, but this bit is pretty exciting, and
I see in the near future, I think in the
next couple of years, more and more franchises who's trought
to use this and you see this tectonic and how
the relationship between Z and so it plays out because
it'll be less about he said, she said, and more
(37:07):
about sort of actual data that you could both work
towards solving.
Speaker 1 (37:11):
It's funny, you said, I was going to interject and
say something similar in that we we live in a
world right now where franchise ors and franchisees often have
a lot of tension between them, and sometimes because we're
not speaking from the same playbook, the franchise sees it
from one perspective. Franchise or sees it from another. But
the one thing that is universal and is not arguable
(37:32):
is the data. Right. So if you're using objection objective
data to give coaching advice or or a behavior change
or behavior modification training, whatever you're trying to do with it,
it's a lot harder for the franchise to push back
on it. And we see that from our perspective on
the selling side, the franchise franchise e performance site on
revenue generation all the time, the FBC's are relieved that
(37:55):
they can look at data and then call and say, hey,
it looks like compared to your peers, blah blah blah. Right,
but being able to amalgamate that data and then have
the AI essentially spit out almost a script for the
FBC to say, look, you need to call Bob about X,
y Z, like these are the three things Bob needs
help with, And this how Bob stacks up against his
peers in region. This how Bob stacks up against his
(38:16):
peers and his age demographic, like how long they've all
been franchise owners? Like for I know for some franchise
systems there are sometimes one or two full time people
trying to generate that data and information so they can
have insights to coach with. This can all be done
very easily with a button click. Now, if like Tashar said,
you get your data lined up so then AI can
(38:38):
read it and interpret it and take action on it,
and then give your fbcs what they need to take
their actions, and don't just a personal touch like you.
We can't lose that because people, as much as AI
is awesome, people still want people to interact with. You
don't get any sort of emotional kick from the or
chemical kick in your brain from AI interactions like you do.
Speaker 2 (39:00):
People absolutely absolutely, and I know I'm repeating myself, but
I'll say these two things again. So you have to
be careful about who can access that data. So you
need that sort of labeling on the data side, and
really well, especially when you have this much data, and
it could be dangerous right like when people can access
everything or you're letting AI access everything, So how do
(39:22):
you get that properly to certain people, to certain roles,
to certain context. And second, when you have the insight, now,
how do you connect that with actions? Think about both
these things.
Speaker 3 (39:33):
If you're just.
Speaker 2 (39:34):
Getting that split out response in chat GPT, are you
able to is that changing anything for you, You're still
going to do the same thing. I'm going to send
a slab to my OPS manager, my OPPS manager is
going to send a text message to my digional manager,
regional manager to the owner and there it's not going anywhere.
So also think of how do you close the loop
(39:55):
by actually bringing actions into the fold and what are
some tools and what's that can help you do that
in the more it's consistent way creating a unified solution.
Speaker 1 (40:05):
It's great feedback. Yeah, because at the end of the day,
all is data. All these insights, they are meaningless if
they don't change behavior, and that's the hard part. So
beautiful and that's where the human personal touch can come
in as well as Hey, here is the insight. This
is what we talked about, Steve, I don't see you've
taken any action on that, Like, how do you know?
You can see it in the data? So I love that.
(40:27):
And that's where coaches can really coach, right, their whole
title is to coach, But you're not doing a lot
of coaching if all you're doing is data analysis and
chasing your tail answering basic questions, which again AI can
do way better than them anyway.
Speaker 3 (40:40):
Yep, absolutely, Well, I.
Speaker 1 (40:43):
Know, Tashar, we're a little tight on time. Now, tell
me any any parting thoughts on where you anticipate people
can see the greatest lift right now from leveraging AI
in their franchise operations. Like if they had to pick
one of the three things we talked about today, where
should somebody start?
Speaker 2 (41:00):
So that's a great question, and I'll start by saying
where not to start? Right? So where not to start
is you're getting excited about AI and everybody is using AI,
and I have this pomo and I just want to
use AI. Right, don't do that unless you're starting with
a proper business outcome, and we have done that. Everybody
has done that. We have to start with the business outcome.
(41:22):
Pick out whatever is the first priority for you. Some
of these are hard to do, some of these are
easier to do. Some of these united you need the
right vendor. Obviously, I want to say, reach out to
us and we can help you do all of this,
but that's that's not the point of this. I would say,
pick out what one, which one is priority one for you,
and then also rank them in order of difficulty, which
(41:44):
means I can do the answering instead of that bit.
But I don't have data, so it's going to be
very difficult for me. I can do the training bit
or it's not a priority for me, but it's easier
to do, right, So do that, create a matrix, see
what is HRN priority and easier to do, and start
with that. I always recommend start experimenting in a very
(42:05):
low cost manner. Today you can do it absolutely for free.
Experiment with a few of your owners and see that's
actually helping you get to the business outcomes. For example,
if I want to automate my support, I can again
create a customer GPT thealthree of my owners. Hey, we're
experimenting with this. Let's try using this for a week
and see if like your questions are getting answered or not.
(42:27):
They're getting answered, and I can seat of e and
put it in a sheet that a number of requests
I was getting before number of requests and getting Now
now you know it's solving the problem and you can
allocate budget to it, which makes sense. So you can say, okay,
I want to scale this to all one hundred locations
and I can now properly create that sort of budget
and what it would mean for me if I can
(42:50):
reach that outcome. So that's that's what I would say,
just to start with the business outcome, not just with AI,
and go step by step manner that a lot of
flex smart people I know DA if they can reach
out to you, if they need help, they can reach
out to me. But yeah, with some of this stuff,
it's easy to get started, like at your home, just
(43:10):
on your computer.
Speaker 1 (43:13):
I agree. That's one of the beauties of what's shifting
in AI. It's kind of like access to being literate.
Back in the day, Like, yeah, nobody had access to
learn how to read, and so only very few people
had access to the data. But it was like that
with AI, only specific people at a massive amount of
money could access AI. And now it's not that way.
Anyone can do it. Yes, I'd love that to Shart.
(43:34):
You've given us a lot of good insights today. If
somebody wants to reach out to you or connect with you,
what's the best way for them to do that.
Speaker 2 (43:41):
The best way is go to delight free dot com,
d E L I G H t R E E
dot com. Reach out to us. We will be available
if you want to reach out directly to me. It's
my first name. To shore p U s h a
r At delight free dot com. I answered almost to
all my emails, So you have any questions, you don't
(44:01):
have to talk about the lightly. Even if you want
to talk about AI and how you set up a
certain experiment, I'm happy to help.
Speaker 1 (44:06):
Yeah, I figured you'd say something like that, Dashar, You're awesome.
Thanks for making time to share some of those insights
with us and get a little bit deeper into reality
within AI and how it can be applied. I sure
appreciate you doing that today.
Speaker 3 (44:19):
Absolutely thanks. Thanks about you. Was be chatting with you
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
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
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!