FULL SHOW NOTES
https://www.microsoftinnovationpodcast.com/565
Unlock the secrets of enterprise data management with insights from Microsoft’s own Julie Koesmarno, a principal product manager in Dataverse. Julie's extraordinary journey from database enthusiast to MVP, and now a key player at Microsoft, will leave you inspired. From her passion for ice cream and travel to her innovative projects within Charles Lamanna's Business Applications Platform Group, Julie’s story is a compelling blend of personal and professional triumphs.
Explore the critical relationship between Copilot AI systems and enterprise data management. Julie sheds light on the concept of 'knowledge' as enterprise data, tackling the challenges and opportunities of integrating Microsoft Graph and Copilot. Learn why robust data housekeeping and security measures are non-negotiable as organizations adopt AI tools, and why grounding AI responses with accurate data is essential for maintaining integrity and governance.
Experience a practical demonstration on data modeling and table creation using Copilot. Julie takes us through setting up an environment, designing data models with natural language, and creating a student accommodation review system. She also shares her journey as a female in tech, emphasizing the importance of a supportive team environment and active listening. Julie's reflections on overcoming challenges and fostering diversity offer invaluable lessons for anyone navigating the tech landscape. Tune in for an episode packed with expert advice and inspiring stories!
If you want to get in touch with me, you can message me here on Linkedin.
Thanks for listening 🚀 - Mark Smith
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Mark Smith (00:01):
Welcome to the Copilot Show, where I interview
Microsoft staff innovating withAI.
I hope you will find thispodcast educational and inspire
you to do more with this greattechnology.
Now let's get on with the show.
In this episode, we'll befocusing on the integration of
custom Copilot experiences withDataverse and Power Apps.
(00:22):
Today's guest is fromWashington in the United States
Well, not originally, but she'lltell you about that.
She works at Microsoft as aprincipal product manager in
Dataverse.
She's also a co-author of SQLServer 2017 Machine Learning
Service with R from way back inthe day.
Shaping Copilot experiencewithin Dataverse, powerapps and
(00:42):
Copilot Studio is one of hermissions.
You can find links to her bio,social media, etc.
In the show notes for thisepisode.
Welcome to the show, Julie.
Julie Koesmarno (00:52):
Awesome.
Thank you so much, Mark.
Thanks for having me.
Mark Smith (00:56):
It's great to have you on the show.
I always love for guests totell a bit about their story
outside of work, and of coursework does encroach over into a
lot of what we do outside ofwork.
But tell me food, family andfun what do they mean to you?
Julie Koesmarno (01:13):
Oh, food, okay, I like ice cream, so I'm going
to talk a little bit about that.
But there's a new one.
It's funny because, since we'retalking about family as well,
so a lot of my family membersare scattered around the world,
some in the UK, some inAustralia, and then here I am in
(01:34):
the States.
From time to time.
I do get you know friendsvisiting.
Some are MVPs, like yourself,right?
So some of my fun memories ofthat having ice cream is when,
like my friend Heidi actually,who is an MVP in the data
platform side, we go to likeshop around this like ice cream
(01:55):
aisle and we look at this MochiBoba ice cream and yeah, nice,
nice.
So that's my favourite oneright now.
That's awesome.
Mark Smith (02:04):
That's awesome.
That's awesome.
And what do you do for funoutside of eating ice cream?
Julie Koesmarno (02:10):
Kind of have to work it off, don't I?
So, yeah, my favorite isactually to be able to do it in.
You know quite a bit for work,but in general way, when I was
(02:32):
younger, I was traveling a lotmore than I do now.
Mark Smith (02:35):
Yeah, yeah, do you have a favorite country?
Julie Koesmarno (02:38):
Oh, that's a tough one.
Well, I cannot not mentionabout my what I consider as my
home country now, which isaustralia.
Well, now that I've lived herefor a long time, so it kind of
gets a bit confusing like whereis my home country?
But I grew up in australia, sothat has to be my favorite,
(02:58):
because most of my familymembers is there.
But if I I were to visitsomewhere else, I would have to
say maybe Holland mostly becausealso for the got a few friends
there really good friends thereand food's pretty good yeah.
Mark Smith (03:20):
Very cool, very cool .
How did you end up being inMicrosoft?
I know you're an MVP at onepoint, but how did your life
path take you to Microsoft, andparticularly even to Microsoft
Redmond, being that there's 14subs around the world that you
could work within, but you'reworking, as they say, for the
mothership?
Julie Koesmarno (03:59):
Yeah, yeah, great point, great question
there.
So I started my career actuallyin the database area.
So know, some of you areprobably out there that have
been in the database worldprobably kind of knew that, but
anyway, so I ended up somehowlanding in the database space
more, which is SQL Server orwhich was SQL Server at the time
, yeah, and then I got awardedMVP through all the community
(04:20):
work, the community work, um,and I figured out like hey, it's
actually really good to, youknow, be able to advocate for
microsoft products and solvingcustomers problems with the, uh,
you know, with the, with themicrosoft spec.
Some of you this is kind ofdating me a little bit some of
you probably remember of the biburger back in the day with the
(04:43):
ss, with the SSRS and all thatkind of stuff like the data we
have in Slack.
So I used to kind of recalllike, oh, that's kind of fun.
I do wonder, like how wouldthey do that?
Like how would Microsoftactually think about coming up
with all these differentfeatures or products?
So I decided one time just toapply and got to meet with a few
(05:06):
Microsoft folks and yeah, andthen I applied and then they let
me, you know, choose which teamI wanted to be part of.
And then I decided, since mybackground was actually
analytics, I decided to join theSQL Server telemetry team.
(05:26):
That was an amazing space tolearn about how SQL Server was
built, because I had to figureout all the different features
that are available out there.
So, yeah, I got to work with alot of telemetry, so it was kind
of fun to work with a lot oftelemetry, so it was kind of fun
.
And then I decided, hey, I wantit to be more of a front-facing
(05:48):
product management sort of role.
So how do I get into that?
And eventually landed inCharles' org Charles' awareness
org, which is BusinessApplications Platform Group, and
I've been very thankful of thatbecause there's a lot more
exposure for trying a lot of newthings.
Like you know, some of thethings that you kind of
(06:10):
mentioned during theintroduction about, you know
Core Pilot.
My team have been able to kindof work on that, so it's very
exciting stuff.
Mark Smith (06:19):
That's awesome and so, specifically, are you well,
what area are you working in?
You know the Power Platform isa big product.
Well, it's many products.
It's an interesting ecosystem.
Are you specifically because ofyour database background in the
Dataverse area of things?
Julie Koesmarno (06:38):
Yeah, great question.
It somehow sort of morphed intobeyond just the database aspect
of it, but essentially justworking with data overall.
Now this is a bit of a funstory.
I recall the first time I hadmy Power Platform, the MPPC, the
(07:00):
Power Platform Conference thevery first time I joined I was
co-presenting with Tim Solenik.
You probably, I think you'vehad him on the show right?
Mark Smith (07:09):
Yes, yes, I have James.
Julie Koesmarno (07:10):
So we're just having a chat and I was saying
like, hey, you didn't realizethat in SQL I could do this
really fast because I know thesyntax of SQL pretty well.
And so since then I was thinkingabout, hmm, maybe the user
experience for, you know,creating tables in general or
bringing data into Dataverse forespecially for enterprise data
(07:33):
or for business applications,could be way improved.
So it's just taking you knowfrom some of my background, as
you know from the pro code, buthow do we introduce it so that
if or when you are low-codedevelopers, or even if you are
pro-code developers, it shouldbe pretty easy for you to create
tables and bring data in?
(07:53):
So it all started from there.
Some of the thinking that I hadat the time was can we use
something like GPT?
This was before the open AI.
You know big announcement onthe chat GPT, so we're exploring
that.
And then, sure enough, withinweeks, that announcement was
there and we were able toleverage a lot more on the large
(08:17):
language models for a number ofthings that we are enabling for
makers.
So I have to say I'm very, veryfortunate to be part of this
journey, being able to enablefor our local makers.
Mark Smith (08:33):
Nice.
So your day-to-day roleinvolves what?
Are you just focused now onCopilot, or are you focused on
Copilot in the context of data,which?
You know, every AI needs datato operate.
How do you explain that?
Julie Koesmarno (08:53):
So the team that I work in, we ensure that
you can bring knowledge easily.
And if you are to use knowledgefor your Copilot, you can do
that easily as well, and if youare to use knowledge for your
co-pilot, you can do that easilyas well.
So, essentially, everythingthat I'm working now is about
knowledge.
Mark Smith (09:10):
Okay, so let's expand the concept of knowledge.
What does knowledge mean to you?
Julie Koesmarno (09:15):
Great question.
So knowledge right now to memeans enterprise data.
Enterprise data could also begeneral knowledge as well, you
know, like what's available inthe public out there.
But for the context of most ofthe time, I'm thinking about
more in the enterprise data sideof things, and it's actually
beyond data first.
(09:35):
So some of the things that weare enabling for within the
Copilot studio where you canground your own custom Copilots
with enterprise data it'scross-collaboration with a lot
of different teams, of course,but one of the things that we're
enabling is not just being ableto allow makers to grant their
custom co-pilots with dataversus enterprise data, but also
(09:58):
some of the other what we callthird-party connectors right,
the co-pilot connectors whichactually technically, under the
hood, is leveraging theMicrosoft Graph connections.
So very exciting space.
Mark Smith (10:13):
So just because I haven't fully got knowledge
nailed down in my head, becauseare you thinking or are you
meaning like, let's say, as anorganization, we have all our
enterprise data and they'll be,you know, distributed across
many systems.
So, you know, from a file shareperspective, we could be on
(10:34):
some HP trim type environment orwe could be in, of course,
share do have a lot ofSharePoint repositories.
You know that data would sit in, but it could be various SQL
(10:55):
systems, it could be, you know,anything from POS systems to ERP
systems.
When you're talking about theorganization's knowledge, are
you talking about all their data?
Julie Koesmarno (11:08):
That's the mission, yeah.
Mark Smith (11:09):
Yeah, awesome.
So here's one of the thingsI've been discussing with
friends lately, which is theminute you light up all that
data to the graph.
Any data sloppiness ispotentially going to be exposed,
right.
Right, and for years peoplehave not known that data was
(11:31):
accessible because they didn'tknow how to query, right?
Because day-to-day you know,individuals are not experienced
queries as in.
The best they could do isprobably do a search engine
search, and most people haveused search engines way below
what their capabilities are.
I remember back in the dayreading a book when I was in web
(11:54):
design, and this was a verylong time ago.
There was a book called GoogleHacks and it was basically
designed to allow you to dosearch querying, you know,
through just the standard searchtool that Google had, and, of
course, you could extract allsorts of things credit card
numbers, passwords, alldifferent types of file formats,
(12:16):
etc.
If you knew how to create areally robust query, you could
access that data.
Now, with the graph theMicrosoft graph coming online in
the context of co-pilots, inthe context of all knowledge of
an organization, there'spotentially a massive amount of
data sloppiness and governancestandards that perhaps haven't
(12:38):
been implemented because noone's known how to find that
information.
And, of course, in a world whereyou can, you know, ask an LLM
that has reasoned over thatinformation.
Ask an LLM that has reasonedover that information, even PII
data, you know, where it mightnot contain the full data set of
an individual, like their name,their date of birth, their
address, but there might befragments of their data all over
(13:01):
the organization.
All of a sudden, llm, brilliantat finding patterns, can all of
a sudden start bringingtogether a lot of information
that in its individual formatswould be meaningless but as a
consolidation would be veryinteresting.
Yeah, what are you seeing inthis space?
(13:24):
Does a co-pilot implementationreally needs to carry out a
security sprint following toactually do some housekeeping
around the data that is nowexposed?
Julie Koesmarno (13:37):
Yeah, a great point.
Actually, that's one of thethings that we have discussed
before.
You're actually bringing afresh perspective from the other
dimension.
You're actually bringing afresh perspective from the other
dimension.
So a lot of the discussionsthat I've been into recently
it's kind of like garbage in,garbage out, Like how do you?
make sure that the data is cleanenough to be consumable.
(13:59):
So that's one aspect, butyou're also adding that
dimension of the security.
So garbage security in, garbagesecurity out.
So definitely those are the twothings that we are considering
as well.
So something for, aspractitioners or, as you know,
customers are adopting this, wehighly recommend to kind of
(14:22):
think through how you organizeyour data before, of course, you
know the whole.
data governance is an importantconcept, but now it's even
making it more important orprevalent to all enterprises.
It's very important to kind oflook into it.
Especially if you're I meanespecially when you're building
(14:44):
apps, you should look into it.
But with Copilot, of course, itmakes it super easy.
So therefore, there's much morekind of care need to be uh
given to, uh to the governanceof data in itself.
Mark Smith (14:58):
Yeah, nice, you talked about grounding your
information, and can you explainwhat you mean by that?
Julie Koesmarno (15:04):
yeah, so with um, with gpt.
So if you are actually usingthe um copilot that when you go
to like Bingcom you can ask youknow the Copilot version, like
you know search stuff right.
So that is typically when theanswer is provided.
It's usually grounded with somedata In most cases when you ask
(15:26):
kind of generic questions likewhere is Gold Coast?
And then it kind of answer likeGold Coast is in Australia.
So it typically sort of likesources and provides citation
hey, you know, this informationcomes from this.
So that's what we mean bygrounding with the data, so that
it's specific to what theco-pilot can answer.
It doesn't use like Bing todayor the co-pilot that you can use
(15:50):
from Bing.
It doesn't actually useenterprise data on your local
machine.
So because you don't set it upthat way or nobody set it up
that way.
So when you build a customco-pilot and this is an
experience that already existedeven prior to build where you
can build a custom co-pilot andyou can say, hey, I just want to
(16:13):
ground it or I want to use itas a data source for the code
palette, just maybe only fromthis website or from this
SharePoint site or from thisfile.
Just making it super easy toillustrate.
Say, if I was to upload a filethat discusses onboarding guide,
then if and I add it to mycustom copilot so that copilot
(16:38):
is grounded with that onboardingguide, so if I were to ask
questions like where's the bestsushi, then it shouldn't answer
that, right, because it's justanswering based on the data that
I've provided it as a maker.
Mark Smith (16:56):
Nice.
So, with the connectors thatyou're talking about, is the
idea that if enterprise data isavailable or knowledge is
available that you will be ablethrough.
Are you more focused on CopilotStudio, like as in custom
Copilots, or are you more in theconcept of M365 Copilot and
(17:20):
bringing the data in and makingthat available to the
organization?
Julie Koesmarno (17:24):
Yeah, great point there.
So the ones that we arereleasing or that we have
released in Build is related tocustom Copilot specifically, and
there is also a Copilotextension that you can do, and
that is usually done via plugins, so we're working on improving
(17:46):
that, so hopefully it's more ofa seamless experience.
So when we talk about co-pilotconnectors in general, it means
it means knowledge, but it alsomeans plugins, which essentially
is connectors or actions, right, so yeah, yeah, and role-based
access control.
Mark Smith (18:03):
Is that all part of the thinking here?
Because you know, from myhistory in dataverse and
pre-dataverse one of our bigselling points to customers was
this you know that we couldsecure down to a field level
whether someone was allowed tosee it edited, updated, that
type.
You know the normal currenttype processes through the ui is
this do those you know whatyou're producing therefore honor
(18:25):
that type of role-based access?
So we don't have people, forexample, say, you know, let's
say, a data set in Dataversethat their role doesn't give
them privilege to.
Julie Koesmarno (18:36):
Yeah, so if you're using Dataverse as your
knowledge to your co-pilot, thenit will honor the role-based
access, the RBAC, for sure.
Excellent.
Yeah and including, even if wetalk about you know the graph
connections.
It will also honor that.
You know whatever data accessthat you may have.
Obviously, you know whateverthat is set up from Microsoft
(18:59):
Graph.
Mark Smith (19:00):
Awesome.
What are you most excited aboutthe next 12 to 24 months in?
What you're working on and whatyou can working on, what you
know and what you can sharearound you know, and why I?
I don't even think 24 monthsprobably nowadays is too far
ahead to be looking rightbecause so much change is
happening in the 12 month window.
We're in what, but what'sexciting you?
Julie Koesmarno (19:22):
I mean it's it's.
It's hard to, uh, you know,predict what the future is like
in terms of technologyperspective.
But I think, um, this, uh, youknow our our journey in CoPilot.
I think it's just begun.
I know that we started it overa year ago, but I think there's
more to come.
Some of the things, like youknow, we discussed about the
(19:42):
data governance that will becomemore important.
I think people are going tocare more and more about their
data.
Yeah, lots of exciting thingsto look forward to.
Mark Smith (19:54):
Nice.
Now an exclusive on the showwe're going to do a demo.
Is that right as in?
We don't normally.
So if you're listening to this,the demo will be over on
YouTube, on my channel.
But, Julie, do you want to goand show us a few things?
Julie Koesmarno (20:11):
Yeah, I actually want to show you a
couple of things, and I'll tryto do the kitchen show, you know
, or the cooking show method,where Nice, let me show you,
like, how you get started.
And then here's one that I kindof created earlier.
So let me start sharing myscreen.
The first one that I wanted totalk about is how, as a maker,
you can get started with datamodeling and inserting or
(20:32):
creating data or creating tablesrather really easily with the
help of Copilot.
If you want to try this at home, and depending on when you're
airing this Mark, there is a fewsettings that you have to do,
because at the moment it's stillin the early preview sort of
stage.
We will make it public previewvery soon when you create a new
(20:54):
environment or if you alreadyhave an environment where it's
in the preview region.
So that means you have to go tothe admin center and look at
your environment settings tomake sure that it's in the
preview region.
You can see this.
Try the New Data Experience.
Now, when you click on the Trynew data experience, either you
can tap in from here or startwith data, or this is actually
(21:17):
to build app.
But if you want to just stop atthe table side.
Then you go to the tables andthen click on this beautiful
button called Create new tables.
Here you'll be presented what Iwould say as kind of like a
blank canvas where you can startdoing some data modeling with
Copilot For the purpose of thediscussion.
(21:38):
I'm just going to start usingthe start with Copilot.
So essentially, using naturallanguage, we'll start designing
or modeling data using Copilotwith the help of Copilot there.
But you could also do similarthings with SharePoint and Excel
file, where you bring in yourexisting data to create new
tables.
I'll go ahead with clicking onthe start with Copilot For the
(22:01):
purpose of this demo.
I'm just going to say studentaccommodation reviews.
So essentially, what I'mproviding here is just like a
prompt hey, I want to create adata model or ERD diagram,
really to like a system to storestudent accommodation reviews.
(22:23):
So I can do some you knowoptimization in terms of like
hey, do you want just the onetable, or multiple tables?
What table size, et cetera,whether to include relationships
or not, and then I'll clickgenerate.
So it'll take a few seconds tocreate.
Mark Smith (22:41):
That's pretty cool because I remember a year,
almost a year and a half ago inCharles Lamanna's session, when
this first came out, you couldonly do a single table.
You couldn't do anything likerelationships or anything like
that.
Julie Koesmarno (22:53):
That is cool to see that that new functionality
there yeah, because my data set, or I guess not that I said um
domain that I'm actually talkabout.
You know what?
I absolutely relate to studentaccommodation reviews.
That's why it comes up with two.
You can add more tables, by theway, after as well.
Like, within the sameexperience, the same canvas,
(23:13):
have you put?
I think some of the ones that Ifind interesting when you ask
your pilot to do is like hotelregistration systems.
Usually it comes up with aboutthree, uh, three tables or so.
So let's take a look here whenyou click on student
accommodation, it gives you somesample data, and likewise, when
you click on accommodationreviews, you get some sample
data as well.
So that way, as you as a maker,as as you're creating or
(23:37):
designing your tables, you cankind of roughly, especially if
you want to create apps afteryou can roughly start testing
out your app with some data,right, because it's usually
pretty helpful to do that Incase you want it to be a bit
more relevant to you, in caseyou want it to be a bit more
relevant to you.
So for me since we talked aboutAustralia earlier, so I could
(23:58):
ask Copilot to help me you know,change this value of CTA to CTE
.
It just doesn't make sense.
So I'm going to say hey, changelocation value in student table
, two cities in Australia, sofingers crossed, hopefully it
(24:21):
works.
Again, just to want toreemphasize this is still in
early preview, so hopefully itworks.
Ta-da, locations all changed toSydney, adelaide, and I can ask
add five more rows in student.
Sometimes I'm not good inspelling, but the good news is
(24:45):
that GPT usually picks it uppretty well.
It's not like ah, you know, youhave a typo here.
We roughly know what you'retalking about.
Mark Smith (24:51):
I found that too.
I found that too Brilliant.
I tell you what I wish I hadthis like 15 years ago, when all
the demos I've done in mycareer and the hassle always of
getting sample data that wasmeaningful.
How quick is this?
Amazing.
Julie Koesmarno (25:06):
Yeah, so when you click on save and exit, then
tables will be created.
So I actually already createdthe table, so I won't click on
the save and exit becauseotherwise I'll create duplicated
tables, which I don't want todo.
But yeah, it's prettystraightforward and you can keep
adding tables as well.
Now the next one that I wantedto do.
(25:26):
So you've seen how, as a maker,I'm getting help from Core
Palette to help me design mydata.
Now the question is what can Ido with it after?
So if you're familiar withPower Apps, like Canvas apps
specifically, then when youcreate these tables and create
an app, you have the wholeCopilot sidecar that you can
start adding.
(25:47):
So that way, you can make yourapp to be Copilot-ready, so to
speak, for your end users.
Now, speaking for end users,now we're talking about how
makers can create their owncustom co-pilot so that their
end users can start askingquestions against this co-pilot.
So you can think of this customco-pilot like as if it's a
(26:11):
perhaps a bot that you want toadd to your own website, or
maybe to Teams as well.
So here's one that I actuallyprepared earlier.
So let me just make the realestate a little bit bigger here.
So I've created a studentaccommodation finder custom code
palette.
It's super easy to do.
(26:32):
If we have time I'll show youthe flow.
But once you create it you cankeep adding knowledge.
So in this case I've alreadyadded a few different knowledge
sources.
If you want to add a new one,you just click on the plus add
knowledge and then you canspecify public website files,
sharepoint Dataverse Fabric iscoming soon or any of the
(26:56):
enterprise data that you haveyour graph connections set up
already by your tenant admin.
So just for simplicity, I'llshow you Dataverse, since we
love Dataverse.
Here you'll see somerecommendation regarding what
table should we use.
Here we go.
(27:17):
So it suggests studentaccommodation.
The way it's suggested, it isbased on the co-pilot name.
So if you give meaningfulco-pilot name, your custom
co-pilot, a meaningful namerather than just co-pilot one,
two, three, usually therecommendation here will be a
little bit more tuned to thatcopilot that you're creating.
(27:38):
So I've already added this.
But just for the sake of demo,let me just show you what that
looks like if you were addingfrom scratch.
You can add multiple tables upto 15, whether related or
unrelated, it doesn't matter.
Under the hood, the system orthe core pilot will figure out
(27:58):
the relationships.
Interestingly, and a bit of, Iguess, lifting the details a bit
more here, we use what we calla natural language to SQL.
So that's our reg or Ritu foraugmented generative or
generated method.
So under the hood, we willtranslate your prompt into SQL,
(28:25):
so your natural language to aSQL statement.
From that perspective, thesystem will just figure out what
would be the relevant tables touse to answer your questions.
So once you're happy with theselection of the tables, you get
to preview it, just to makesure that these are the right
tables.
This preview is just showingyou a few columns, it won't show
(28:46):
the entire columns.
Once you're happy with it, thenyou click Next and then you can
also provide things likesynonyms.
Here, as an example, you couldsay let's go to the student
accommodation one.
Here, for example,accommodation name.
Maybe you want to add thingslike dorm name or whatever
(29:08):
colloquial names that areapplicable to you or to your
enterprise.
We can also add glossary as anexample, colloquial kind of
names that are applicable to youor to your enterprise.
We can also add glossary.
As an example, we can say hey,akomu, sorry, not accommodation.
Let's say housing is the sameas same as student accommodation
or maybe hostel or somethingthen you click back and then you
(29:33):
click Add.
Once you click Add so I won'tadd it for now but once you
click Add, it will be added tothe screen here where you can
also navigate to knowledge.
You can edit it further.
This is the one that I'vecreated before.
Previews, synonyms, glossary,it's all the same thing.
(29:56):
Then this is where I can startasking questions which is
earlier asked, like hey, whichDOM has rating of more than four
?
We can ask it again.
Oops, not represented.
Let me just click allow.
Oh, my control, see, controlfee works.
There you go.
(30:18):
Yeah, so it can start askingquestions and it will provide
citation as well.
So in this instance it will.
It will take a little bit, Ithink, to answer, but in this
instance where, oh, there you go, it comes up with the answers
already.
In the instance where, let mejust go back to knowledge.
In the instance where, oh,there you go, it comes up with
the answers already.
In the instance where, let mejust go back to knowledge.
In the instance where I askquestions related to who wrote a
(30:39):
book about, you know,australia's dangerous snakes,
this question is best sourcedfrom the PDF file that I've
provided, then it also providesthat citation as well.
So if you click on it, you getto see the preview of that, a
simple preview of that PDF file.
Yeah, so this is how you testyour core palette and eventually
(31:03):
, once you do the publishingwhich I haven't done for this
environment yet in case you wantto do publishing sorry, let me
just go to channels rather thanpublish.
This is an area where you cancontrol or you can manage the
channels that you want to enablethis co-pilot for.
So one of them is MicrosoftTeams.
So this is one that I'vecreated earlier.
(31:25):
So a similar co-pilot that I'vepublished to Teams, where it
actually just grounded with theaccounts table, I can ask
questions like where'sAdventureWorks located?
It's okay in Santa Cruz.
So this is how we think thefuture will be, where you can
now start asking co-pilotquestions, not just from, for
(31:49):
example, just from your websiteas a chatbot or so so, but more
in the productivity tools likemicrosoft teams.
Mark Smith (31:57):
So there's a lot of powerful things that you can
leverage from co-pilot studio Isee that, um, audio queries are
going to become the norm, rightas in.
Like, you won't even have tonecessarily go through an app
interface, you'll just be ableto query and get answers.
It's an awesome change for me.
You know, dyslexic typingspelling is always an issue I
(32:20):
don't have.
I have a lot less issue withthat now in a co-pilot world, so
it's definitely an enablingtype tool.
Before I let you go, I wrote apost in 2018, 2019, and was
controversial.
Um, and it's something that Ilike to do every now and again,
(32:41):
I like to be controversial toget a discussion going.
It was around a lot of peoplewere complaining around things
like diversity and inclusion andthings like this that you know,
in my case, if you're amiddle-aged white guy in tech,
they don't want you to front anyevents anymore or anything
because you're not, you know,diverse enough.
(33:01):
And so I wrote something with aheadline to that effect and, of
course, a lot of people onlyread headlines they don't read
the blog post and, of course,all the people that read the
headline agreed with me were allthose people that were anti
this type of sentiment.
But as I go down there, as Idrilled into the topic.
I said the importance of whensomebody looks up to somebody
(33:23):
presenting in tech leadershipand of course, I'm only focused
on tech because that's what I'min it's important that they see
somebody like me and so,therefore, if it's a teenage
girl looking, they need to seefemale role models.
If it's any across theinclusion or diversity landscape
, people need to see role modelsthat they go.
You know what.
(33:43):
I identify with them becausethey're similar to me.
I can see, you know, andalthough middle-aged white guys
have had the stage for manyyears, it is time for change.
What are your thoughts aroundthat?
Women in tech what are yourthoughts?
Julie Koesmarno (34:01):
Yeah, that's very interesting.
So here's what I will say, witha little bit of, I guess,
caveat, like I cannot representthe entire women in tech, but I
can share some of myperspectives, which is, I would
probably say, to a certaindegree, unique on its own, but
maybe some of the learnings canbe useful for somebody out there
(34:22):
.
So a bit of background.
I do, or did have the privilegeof having quite strong role
models.
So, for example, like my mom,mom owns a business and it's in
the hardware sort of.
When we talk about hardwareswe're talking about like nuts
and bolts, not software,software, not computer or
(34:42):
anything like that.
So it was pretty uh interestingto see.
So it's sort of like, oh, momcan do that.
So I think I could probablykind of get into more on the
technical side of it as well,like if needed, and there's a
lot of business thinking as wellas math thinking around that.
So that's one aspect of me.
(35:03):
The other aspect of me like youknow how I grew up that I also
consider as privileged too, isthat I did grow up in Indonesia
when I was much younger andstudying computer science was
mandatory.
Everybody got to learn it and Ijust happened to love it.
So I usually kind of like acethe class.
(35:24):
So, coming to a more Westerncountry so as an example,
australia I found myself alittle bit dumbfounded in in
which I chose my elective, mycomputer science, to be my
elective, and I was the onlyfemale.
Also, non-english English is mysecond language too, yes, so I
was like, oh wow, like the worldis very different.
(35:48):
Um, so I had some of the thingsthat kind of helped me think a
little bit more that the worldis just outside of me, that
there are a few different sortof like diverse perspectives.
So what I from personally, frommyself, I try to, I strive to
(36:09):
give different perspectives toall, at least my perspective to
other folks, and also to welcomeothers as well.
Now there are lots of articles,discussions out there where
women are struggling in thisspace.
I'm not discounting that,definitely it's there.
So I wanted to give a bit ofdifferent perspective what it
(36:32):
feels like to be supported, andjust to a bit of a disclaimer,
what it feels like to besupported, and just to a bit of
a disclaimer as well.
It's not like all through mycareer I'm always supported.
That's not the case.
It's just that there are a fewpoints in the career that in my
career that I noticed oh, I knowwhen I think I feel at my best,
where I feel supported, andthat looks something like this
(36:55):
Usually the discussion withinthe team everybody feels like
it's okay to make mistakes.
So we, and it's okay that we areon the basis of learning, we
didn't know, we didn't knowabout something, and then we use
it as a learning experience.
Sometimes it's to do withtechnical stuff like oh, I'm
(37:17):
sorry, I'm kind of new to this.
Can you help me understand X, yand Z?
It could be about co-pilotknowledge or whatever, or what
RAG means or what LLM means, andthen being able to do kind of
like active listening andlearning from there.
But in some cases it might be abehavioral thing.
So if somebody said something alittle bit, maybe a little bit
(37:37):
hurtful, if you can and this isvery, I would say, depending on
the power dynamics and yoursituation right, if you're in a
situation where you can feelsafe then you could say hey, you
know, can we talk a little bit?
I think that the stuff that youmentioned before could be
interpreted differently forother people, so how do we think
(38:01):
about rephrasing that?
Also, what's the intent behindit?
Another thing that I wanted tobring up is just because it's
not your intention.
So if you say hurtful thingsit's not your intention, though,
but it doesn't mean that it'sless hurt for the recipient.
Mark Smith (38:17):
A hundred percent.
Julie Koesmarno (38:18):
Overall overall .
A couple of things that I wouldsay I draw from this experience
is that when I'm in the teamsetting, when it's not zero sum
game, so when we do we playwin-win, I think it's, um, not
just from women perspective, butI think from the team all
(38:40):
around.
We feel like we are respected,um, we can be put into a
position that, hey, I'maccountable for this, my
teammates are, know, making meaccountable for this too, and
you actually, you know, becomeyour better self too.
So those are some of the thingsthat I've learned so far.
(39:01):
Where I started picking, youknow, some of the behavior not
behaviors, but some of thesymptoms I don't know what
characteristics, characteristicsof when I feel a bit more
supported.
Um, then some other experiences.
So, and I have to say, where Iam right now, I happen to get to
(39:22):
witness that a lot.
So very, very thankful of ofthat too yeah, that is so good.
Mark Smith (39:29):
That is so good and I love that story, particularly
starting with your mom andhaving her in business.
That's phenomenal, Julie, it'sbeen awesome to have you on the
show.
Thank you, I look forward tohaving you back again in the
future.
Julie Koesmarno (39:43):
Awesome.
Thank you for having me.
Mark Smith (39:46):
Hey, thanks for listening.
I'm your host, mark Smith,otherwise known as the NZ365 guy
.
Is there a guest you would liketo see on the show from
Microsoft?
Please message me on LinkedInand I'll see what I can do.
Final question for you how willyou create with Copilot today,
ka kite?
Welcome to the Copilot Show, where I interview
Microsoft staff innovating withAI.
I hope you will find thispodcast educational and inspire
you to do more with this greattechnology.
Now let's get on with the show.
In this episode, we'll befocusing on the integration of
custom Copilot experiences withDataverse and Power Apps.
(00:22):
Today's guest is fromWashington in the United States
Well, not originally, but she'lltell you about that.
She works at Microsoft as aprincipal product manager in
Dataverse.
She's also a co-author of SQLServer 2017 Machine Learning
Service with R from way back inthe day.
Shaping Copilot experiencewithin Dataverse, powerapps and
(00:42):
Copilot Studio is one of hermissions.
You can find links to her bio,social media, etc.
In the show notes for thisepisode.
Welcome to the show, Julie.
Julie Koesmarno (00:52):
Awesome.
Thank you so much, Mark.
Thanks for having me.
Mark Smith (00:56):
It's great to have you on the show.
I always love for guests totell a bit about their story
outside of work, and of coursework does encroach over into a
lot of what we do outside ofwork.
But tell me food, family andfun what do they mean to you?
Julie Koesmarno (01:13):
Oh, food, okay, I like ice cream, so I'm going
to talk a little bit about that.
But there's a new one.
It's funny because, since we'retalking about family as well,
so a lot of my family membersare scattered around the world,
some in the UK, some inAustralia, and then here I am in
(01:34):
the States.
From time to time.
I do get you know friendsvisiting.
Some are MVPs, like yourself,right?
So some of my fun memories ofthat having ice cream is when,
like my friend Heidi actually,who is an MVP in the data
platform side, we go to likeshop around this like ice cream
(01:55):
aisle and we look at this MochiBoba ice cream and yeah, nice,
nice.
So that's my favourite oneright now.
That's awesome.
Mark Smith (02:04):
That's awesome.
That's awesome.
And what do you do for funoutside of eating ice cream?
Julie Koesmarno (02:10):
Kind of have to work it off, don't I?
So, yeah, my favorite isactually to be able to do it in.
You know quite a bit for work,but in general way, when I was
(02:32):
younger, I was traveling a lotmore than I do now.
Mark Smith (02:35):
Yeah, yeah, do you have a favorite country?
Julie Koesmarno (02:38):
Oh, that's a tough one.
Well, I cannot not mentionabout my what I consider as my
home country now, which isaustralia.
Well, now that I've lived herefor a long time, so it kind of
gets a bit confusing like whereis my home country?
But I grew up in australia, sothat has to be my favorite,
(02:58):
because most of my familymembers is there.
But if I I were to visitsomewhere else, I would have to
say maybe Holland mostly becausealso for the got a few friends
there really good friends thereand food's pretty good yeah.
Mark Smith (03:20):
Very cool, very cool .
How did you end up being inMicrosoft?
I know you're an MVP at onepoint, but how did your life
path take you to Microsoft, andparticularly even to Microsoft
Redmond, being that there's 14subs around the world that you
could work within, but you'reworking, as they say, for the
mothership?
Julie Koesmarno (03:59):
Yeah, yeah, great point, great question
there.
So I started my career actuallyin the database area.
So know, some of you areprobably out there that have
been in the database worldprobably kind of knew that, but
anyway, so I ended up somehowlanding in the database space
more, which is SQL Server orwhich was SQL Server at the time
, yeah, and then I got awardedMVP through all the community
(04:20):
work, the community work, um,and I figured out like hey, it's
actually really good to, youknow, be able to advocate for
microsoft products and solvingcustomers problems with the, uh,
you know, with the, with themicrosoft spec.
Some of you this is kind ofdating me a little bit some of
you probably remember of the biburger back in the day with the
(04:43):
ss, with the SSRS and all thatkind of stuff like the data we
have in Slack.
So I used to kind of recalllike, oh, that's kind of fun.
I do wonder, like how wouldthey do that?
Like how would Microsoftactually think about coming up
with all these differentfeatures or products?
So I decided one time just toapply and got to meet with a few
(05:06):
Microsoft folks and yeah, andthen I applied and then they let
me, you know, choose which teamI wanted to be part of.
And then I decided, since mybackground was actually
analytics, I decided to join theSQL Server telemetry team.
(05:26):
That was an amazing space tolearn about how SQL Server was
built, because I had to figureout all the different features
that are available out there.
So, yeah, I got to work with alot of telemetry, so it was kind
of fun to work with a lot oftelemetry, so it was kind of fun
.
And then I decided, hey, I wantit to be more of a front-facing
(05:48):
product management sort of role.
So how do I get into that?
And eventually landed inCharles' org Charles' awareness
org, which is BusinessApplications Platform Group, and
I've been very thankful of thatbecause there's a lot more
exposure for trying a lot of newthings.
Like you know, some of thethings that you kind of
(06:10):
mentioned during theintroduction about, you know
Core Pilot.
My team have been able to kindof work on that, so it's very
exciting stuff.
Mark Smith (06:19):
That's awesome and so, specifically, are you well,
what area are you working in?
You know the Power Platform isa big product.
Well, it's many products.
It's an interesting ecosystem.
Are you specifically because ofyour database background in the
Dataverse area of things?
Julie Koesmarno (06:38):
Yeah, great question.
It somehow sort of morphed intobeyond just the database aspect
of it, but essentially justworking with data overall.
Now this is a bit of a funstory.
I recall the first time I hadmy Power Platform, the MPPC, the
(07:00):
Power Platform Conference thevery first time I joined I was
co-presenting with Tim Solenik.
You probably, I think you'vehad him on the show right?
Mark Smith (07:09):
Yes, yes, I have James.
Julie Koesmarno (07:10):
So we're just having a chat and I was saying
like, hey, you didn't realizethat in SQL I could do this
really fast because I know thesyntax of SQL pretty well.
And so since then I was thinkingabout, hmm, maybe the user
experience for, you know,creating tables in general or
bringing data into Dataverse forespecially for enterprise data
(07:33):
or for business applications,could be way improved.
So it's just taking you knowfrom some of my background, as
you know from the pro code, buthow do we introduce it so that
if or when you are low-codedevelopers, or even if you are
pro-code developers, it shouldbe pretty easy for you to create
tables and bring data in?
(07:53):
So it all started from there.
Some of the thinking that I hadat the time was can we use
something like GPT?
This was before the open AI.
You know big announcement onthe chat GPT, so we're exploring
that.
And then, sure enough, withinweeks, that announcement was
there and we were able toleverage a lot more on the large
(08:17):
language models for a number ofthings that we are enabling for
makers.
So I have to say I'm very, veryfortunate to be part of this
journey, being able to enablefor our local makers.
Mark Smith (08:33):
Nice.
So your day-to-day roleinvolves what?
Are you just focused now onCopilot, or are you focused on
Copilot in the context of data,which?
You know, every AI needs datato operate.
How do you explain that?
Julie Koesmarno (08:53):
So the team that I work in, we ensure that
you can bring knowledge easily.
And if you are to use knowledgefor your Copilot, you can do
that easily as well, and if youare to use knowledge for your
co-pilot, you can do that easilyas well.
So, essentially, everythingthat I'm working now is about
knowledge.
Mark Smith (09:10):
Okay, so let's expand the concept of knowledge.
What does knowledge mean to you?
Julie Koesmarno (09:15):
Great question.
So knowledge right now to memeans enterprise data.
Enterprise data could also begeneral knowledge as well, you
know, like what's available inthe public out there.
But for the context of most ofthe time, I'm thinking about
more in the enterprise data sideof things, and it's actually
beyond data first.
(09:35):
So some of the things that weare enabling for within the
Copilot studio where you canground your own custom Copilots
with enterprise data it'scross-collaboration with a lot
of different teams, of course,but one of the things that we're
enabling is not just being ableto allow makers to grant their
custom co-pilots with dataversus enterprise data, but also
(09:58):
some of the other what we callthird-party connectors right,
the co-pilot connectors whichactually technically, under the
hood, is leveraging theMicrosoft Graph connections.
So very exciting space.
Mark Smith (10:13):
So just because I haven't fully got knowledge
nailed down in my head, becauseare you thinking or are you
meaning like, let's say, as anorganization, we have all our
enterprise data and they'll be,you know, distributed across
many systems.
So, you know, from a file shareperspective, we could be on
(10:34):
some HP trim type environment orwe could be in, of course,
share do have a lot ofSharePoint repositories.
You know that data would sit in, but it could be various SQL
(10:55):
systems, it could be, you know,anything from POS systems to ERP
systems.
When you're talking about theorganization's knowledge, are
you talking about all their data?
Julie Koesmarno (11:08):
That's the mission, yeah.
Mark Smith (11:09):
Yeah, awesome.
So here's one of the thingsI've been discussing with
friends lately, which is theminute you light up all that
data to the graph.
Any data sloppiness ispotentially going to be exposed,
right.
Right, and for years peoplehave not known that data was
(11:31):
accessible because they didn'tknow how to query, right?
Because day-to-day you know,individuals are not experienced
queries as in.
The best they could do isprobably do a search engine
search, and most people haveused search engines way below
what their capabilities are.
I remember back in the dayreading a book when I was in web
(11:54):
design, and this was a verylong time ago.
There was a book called GoogleHacks and it was basically
designed to allow you to dosearch querying, you know,
through just the standard searchtool that Google had, and, of
course, you could extract allsorts of things credit card
numbers, passwords, alldifferent types of file formats,
(12:16):
etc.
If you knew how to create areally robust query, you could
access that data.
Now, with the graph theMicrosoft graph coming online in
the context of co-pilots, inthe context of all knowledge of
an organization, there'spotentially a massive amount of
data sloppiness and governancestandards that perhaps haven't
(12:38):
been implemented because noone's known how to find that
information.
And, of course, in a world whereyou can, you know, ask an LLM
that has reasoned over thatinformation.
Ask an LLM that has reasonedover that information, even PII
data, you know, where it mightnot contain the full data set of
an individual, like their name,their date of birth, their
address, but there might befragments of their data all over
(13:01):
the organization.
All of a sudden, llm, brilliantat finding patterns, can all of
a sudden start bringingtogether a lot of information
that in its individual formatswould be meaningless but as a
consolidation would be veryinteresting.
Yeah, what are you seeing inthis space?
(13:24):
Does a co-pilot implementationreally needs to carry out a
security sprint following toactually do some housekeeping
around the data that is nowexposed?
Julie Koesmarno (13:37):
Yeah, a great point.
Actually, that's one of thethings that we have discussed
before.
You're actually bringing afresh perspective from the other
dimension.
You're actually bringing afresh perspective from the other
dimension.
So a lot of the discussionsthat I've been into recently
it's kind of like garbage in,garbage out, Like how do you?
make sure that the data is cleanenough to be consumable.
(13:59):
So that's one aspect, butyou're also adding that
dimension of the security.
So garbage security in, garbagesecurity out.
So definitely those are the twothings that we are considering
as well.
So something for, aspractitioners or, as you know,
customers are adopting this, wehighly recommend to kind of
(14:22):
think through how you organizeyour data before, of course, you
know the whole.
data governance is an importantconcept, but now it's even
making it more important orprevalent to all enterprises.
It's very important to kind oflook into it.
Especially if you're I meanespecially when you're building
(14:44):
apps, you should look into it.
But with Copilot, of course, itmakes it super easy.
So therefore, there's much morekind of care need to be uh
given to, uh to the governanceof data in itself.
Mark Smith (14:58):
Yeah, nice, you talked about grounding your
information, and can you explainwhat you mean by that?
Julie Koesmarno (15:04):
yeah, so with um, with gpt.
So if you are actually usingthe um copilot that when you go
to like Bingcom you can ask youknow the Copilot version, like
you know search stuff right.
So that is typically when theanswer is provided.
It's usually grounded with somedata In most cases when you ask
(15:26):
kind of generic questions likewhere is Gold Coast?
And then it kind of answer likeGold Coast is in Australia.
So it typically sort of likesources and provides citation
hey, you know, this informationcomes from this.
So that's what we mean bygrounding with the data, so that
it's specific to what theco-pilot can answer.
It doesn't use like Bing todayor the co-pilot that you can use
(15:50):
from Bing.
It doesn't actually useenterprise data on your local
machine.
So because you don't set it upthat way or nobody set it up
that way.
So when you build a customco-pilot and this is an
experience that already existedeven prior to build where you
can build a custom co-pilot andyou can say, hey, I just want to
(16:13):
ground it or I want to use itas a data source for the code
palette, just maybe only fromthis website or from this
SharePoint site or from thisfile.
Just making it super easy toillustrate.
Say, if I was to upload a filethat discusses onboarding guide,
then if and I add it to mycustom copilot so that copilot
(16:38):
is grounded with that onboardingguide, so if I were to ask
questions like where's the bestsushi, then it shouldn't answer
that, right, because it's justanswering based on the data that
I've provided it as a maker.
Mark Smith (16:56):
Nice.
So, with the connectors thatyou're talking about, is the
idea that if enterprise data isavailable or knowledge is
available that you will be ablethrough.
Are you more focused on CopilotStudio, like as in custom
Copilots, or are you more in theconcept of M365 Copilot and
(17:20):
bringing the data in and makingthat available to the
organization?
Julie Koesmarno (17:24):
Yeah, great point there.
So the ones that we arereleasing or that we have
released in Build is related tocustom Copilot specifically, and
there is also a Copilotextension that you can do, and
that is usually done via plugins, so we're working on improving
(17:46):
that, so hopefully it's more ofa seamless experience.
So when we talk about co-pilotconnectors in general, it means
it means knowledge, but it alsomeans plugins, which essentially
is connectors or actions, right, so yeah, yeah, and role-based
access control.
Mark Smith (18:03):
Is that all part of the thinking here?
Because you know, from myhistory in dataverse and
pre-dataverse one of our bigselling points to customers was
this you know that we couldsecure down to a field level
whether someone was allowed tosee it edited, updated, that
type.
You know the normal currenttype processes through the ui is
this do those you know whatyou're producing therefore honor
(18:25):
that type of role-based access?
So we don't have people, forexample, say, you know, let's
say, a data set in Dataversethat their role doesn't give
them privilege to.
Julie Koesmarno (18:36):
Yeah, so if you're using Dataverse as your
knowledge to your co-pilot, thenit will honor the role-based
access, the RBAC, for sure.
Excellent.
Yeah and including, even if wetalk about you know the graph
connections.
It will also honor that.
You know whatever data accessthat you may have.
Obviously, you know whateverthat is set up from Microsoft
(18:59):
Graph.
Mark Smith (19:00):
Awesome.
What are you most excited aboutthe next 12 to 24 months in?
What you're working on and whatyou can working on, what you
know and what you can sharearound you know, and why I?
I don't even think 24 monthsprobably nowadays is too far
ahead to be looking rightbecause so much change is
happening in the 12 month window.
We're in what, but what'sexciting you?
Julie Koesmarno (19:22):
I mean it's it's.
It's hard to, uh, you know,predict what the future is like
in terms of technologyperspective.
But I think, um, this, uh, youknow our our journey in CoPilot.
I think it's just begun.
I know that we started it overa year ago, but I think there's
more to come.
Some of the things, like youknow, we discussed about the
(19:42):
data governance that will becomemore important.
I think people are going tocare more and more about their
data.
Yeah, lots of exciting thingsto look forward to.
Mark Smith (19:54):
Nice.
Now an exclusive on the showwe're going to do a demo.
Is that right as in?
We don't normally.
So if you're listening to this,the demo will be over on
YouTube, on my channel.
But, Julie, do you want to goand show us a few things?
Julie Koesmarno (20:11):
Yeah, I actually want to show you a
couple of things, and I'll tryto do the kitchen show, you know
, or the cooking show method,where Nice, let me show you,
like, how you get started.
And then here's one that I kindof created earlier.
So let me start sharing myscreen.
The first one that I wanted totalk about is how, as a maker,
you can get started with datamodeling and inserting or
(20:32):
creating data or creating tablesrather really easily with the
help of Copilot.
If you want to try this at home, and depending on when you're
airing this Mark, there is a fewsettings that you have to do,
because at the moment it's stillin the early preview sort of
stage.
We will make it public previewvery soon when you create a new
(20:54):
environment or if you alreadyhave an environment where it's
in the preview region.
So that means you have to go tothe admin center and look at
your environment settings tomake sure that it's in the
preview region.
You can see this.
Try the New Data Experience.
Now, when you click on the Trynew data experience, either you
can tap in from here or startwith data, or this is actually
(21:17):
to build app.
But if you want to just stop atthe table side.
Then you go to the tables andthen click on this beautiful
button called Create new tables.
Here you'll be presented what Iwould say as kind of like a
blank canvas where you can startdoing some data modeling with
Copilot For the purpose of thediscussion.
(21:38):
I'm just going to start usingthe start with Copilot.
So essentially, using naturallanguage, we'll start designing
or modeling data using Copilotwith the help of Copilot there.
But you could also do similarthings with SharePoint and Excel
file, where you bring in yourexisting data to create new
tables.
I'll go ahead with clicking onthe start with Copilot For the
(22:01):
purpose of this demo.
I'm just going to say studentaccommodation reviews.
So essentially, what I'mproviding here is just like a
prompt hey, I want to create adata model or ERD diagram,
really to like a system to storestudent accommodation reviews.
(22:23):
So I can do some you knowoptimization in terms of like
hey, do you want just the onetable, or multiple tables?
What table size, et cetera,whether to include relationships
or not, and then I'll clickgenerate.
So it'll take a few seconds tocreate.
Mark Smith (22:41):
That's pretty cool because I remember a year,
almost a year and a half ago inCharles Lamanna's session, when
this first came out, you couldonly do a single table.
You couldn't do anything likerelationships or anything like
that.
Julie Koesmarno (22:53):
That is cool to see that that new functionality
there yeah, because my data set, or I guess not that I said um
domain that I'm actually talkabout.
You know what?
I absolutely relate to studentaccommodation reviews.
That's why it comes up with two.
You can add more tables, by theway, after as well.
Like, within the sameexperience, the same canvas,
(23:13):
have you put?
I think some of the ones that Ifind interesting when you ask
your pilot to do is like hotelregistration systems.
Usually it comes up with aboutthree, uh, three tables or so.
So let's take a look here whenyou click on student
accommodation, it gives you somesample data, and likewise, when
you click on accommodationreviews, you get some sample
data as well.
So that way, as you as a maker,as as you're creating or
(23:37):
designing your tables, you cankind of roughly, especially if
you want to create apps afteryou can roughly start testing
out your app with some data,right, because it's usually
pretty helpful to do that Incase you want it to be a bit
more relevant to you, in caseyou want it to be a bit more
relevant to you.
So for me since we talked aboutAustralia earlier, so I could
(23:58):
ask Copilot to help me you know,change this value of CTA to CTE
.
It just doesn't make sense.
So I'm going to say hey, changelocation value in student table
, two cities in Australia, sofingers crossed, hopefully it
(24:21):
works.
Again, just to want toreemphasize this is still in
early preview, so hopefully itworks.
Ta-da, locations all changed toSydney, adelaide, and I can ask
add five more rows in student.
Sometimes I'm not good inspelling, but the good news is
(24:45):
that GPT usually picks it uppretty well.
It's not like ah, you know, youhave a typo here.
We roughly know what you'retalking about.
Mark Smith (24:51):
I found that too.
I found that too Brilliant.
I tell you what I wish I hadthis like 15 years ago, when all
the demos I've done in mycareer and the hassle always of
getting sample data that wasmeaningful.
How quick is this?
Amazing.
Julie Koesmarno (25:06):
Yeah, so when you click on save and exit, then
tables will be created.
So I actually already createdthe table, so I won't click on
the save and exit becauseotherwise I'll create duplicated
tables, which I don't want todo.
But yeah, it's prettystraightforward and you can keep
adding tables as well.
Now the next one that I wantedto do.
(25:26):
So you've seen how, as a maker,I'm getting help from Core
Palette to help me design mydata.
Now the question is what can Ido with it after?
So if you're familiar withPower Apps, like Canvas apps
specifically, then when youcreate these tables and create
an app, you have the wholeCopilot sidecar that you can
start adding.
(25:47):
So that way, you can make yourapp to be Copilot-ready, so to
speak, for your end users.
Now, speaking for end users,now we're talking about how
makers can create their owncustom co-pilot so that their
end users can start askingquestions against this co-pilot.
So you can think of this customco-pilot like as if it's a
(26:11):
perhaps a bot that you want toadd to your own website, or
maybe to Teams as well.
So here's one that I actuallyprepared earlier.
So let me just make the realestate a little bit bigger here.
So I've created a studentaccommodation finder custom code
palette.
It's super easy to do.
(26:32):
If we have time I'll show youthe flow.
But once you create it you cankeep adding knowledge.
So in this case I've alreadyadded a few different knowledge
sources.
If you want to add a new one,you just click on the plus add
knowledge and then you canspecify public website files,
sharepoint Dataverse Fabric iscoming soon or any of the
(26:56):
enterprise data that you haveyour graph connections set up
already by your tenant admin.
So just for simplicity, I'llshow you Dataverse, since we
love Dataverse.
Here you'll see somerecommendation regarding what
table should we use.
Here we go.
(27:17):
So it suggests studentaccommodation.
The way it's suggested, it isbased on the co-pilot name.
So if you give meaningfulco-pilot name, your custom
co-pilot, a meaningful namerather than just co-pilot one,
two, three, usually therecommendation here will be a
little bit more tuned to thatcopilot that you're creating.
(27:38):
So I've already added this.
But just for the sake of demo,let me just show you what that
looks like if you were addingfrom scratch.
You can add multiple tables upto 15, whether related or
unrelated, it doesn't matter.
Under the hood, the system orthe core pilot will figure out
(27:58):
the relationships.
Interestingly, and a bit of, Iguess, lifting the details a bit
more here, we use what we calla natural language to SQL.
So that's our reg or Ritu foraugmented generative or
generated method.
So under the hood, we willtranslate your prompt into SQL,
(28:25):
so your natural language to aSQL statement.
From that perspective, thesystem will just figure out what
would be the relevant tables touse to answer your questions.
So once you're happy with theselection of the tables, you get
to preview it, just to makesure that these are the right
tables.
This preview is just showingyou a few columns, it won't show
(28:46):
the entire columns.
Once you're happy with it, thenyou click Next and then you can
also provide things likesynonyms.
Here, as an example, you couldsay let's go to the student
accommodation one.
Here, for example,accommodation name.
Maybe you want to add thingslike dorm name or whatever
(29:08):
colloquial names that areapplicable to you or to your
enterprise.
We can also add glossary as anexample, colloquial kind of
names that are applicable to youor to your enterprise.
We can also add glossary.
As an example, we can say hey,akomu, sorry, not accommodation.
Let's say housing is the sameas same as student accommodation
or maybe hostel or somethingthen you click back and then you
(29:33):
click Add.
Once you click Add so I won'tadd it for now but once you
click Add, it will be added tothe screen here where you can
also navigate to knowledge.
You can edit it further.
This is the one that I'vecreated before.
Previews, synonyms, glossary,it's all the same thing.
(29:56):
Then this is where I can startasking questions which is
earlier asked, like hey, whichDOM has rating of more than four
?
We can ask it again.
Oops, not represented.
Let me just click allow.
Oh, my control, see, controlfee works.
There you go.
(30:18):
Yeah, so it can start askingquestions and it will provide
citation as well.
So in this instance it will.
It will take a little bit, Ithink, to answer, but in this
instance where, oh, there you go, it comes up with the answers
already.
In the instance where, let mejust go back to knowledge.
In the instance where, oh,there you go, it comes up with
the answers already.
In the instance where, let mejust go back to knowledge.
In the instance where I askquestions related to who wrote a
(30:39):
book about, you know,australia's dangerous snakes,
this question is best sourcedfrom the PDF file that I've
provided, then it also providesthat citation as well.
So if you click on it, you getto see the preview of that, a
simple preview of that PDF file.
Yeah, so this is how you testyour core palette and eventually
(31:03):
, once you do the publishingwhich I haven't done for this
environment yet in case you wantto do publishing sorry, let me
just go to channels rather thanpublish.
This is an area where you cancontrol or you can manage the
channels that you want to enablethis co-pilot for.
So one of them is MicrosoftTeams.
So this is one that I'vecreated earlier.
(31:25):
So a similar co-pilot that I'vepublished to Teams, where it
actually just grounded with theaccounts table, I can ask
questions like where'sAdventureWorks located?
It's okay in Santa Cruz.
So this is how we think thefuture will be, where you can
now start asking co-pilotquestions, not just from, for
(31:49):
example, just from your websiteas a chatbot or so so, but more
in the productivity tools likemicrosoft teams.
Mark Smith (31:57):
So there's a lot of powerful things that you can
leverage from co-pilot studio Isee that, um, audio queries are
going to become the norm, rightas in.
Like, you won't even have tonecessarily go through an app
interface, you'll just be ableto query and get answers.
It's an awesome change for me.
You know, dyslexic typingspelling is always an issue I
(32:20):
don't have.
I have a lot less issue withthat now in a co-pilot world, so
it's definitely an enablingtype tool.
Before I let you go, I wrote apost in 2018, 2019, and was
controversial.
Um, and it's something that Ilike to do every now and again,
(32:41):
I like to be controversial toget a discussion going.
It was around a lot of peoplewere complaining around things
like diversity and inclusion andthings like this that you know,
in my case, if you're amiddle-aged white guy in tech,
they don't want you to front anyevents anymore or anything
because you're not, you know,diverse enough.
(33:01):
And so I wrote something with aheadline to that effect and, of
course, a lot of people onlyread headlines they don't read
the blog post and, of course,all the people that read the
headline agreed with me were allthose people that were anti
this type of sentiment.
But as I go down there, as Idrilled into the topic.
I said the importance of whensomebody looks up to somebody
(33:23):
presenting in tech leadershipand of course, I'm only focused
on tech because that's what I'min it's important that they see
somebody like me and so,therefore, if it's a teenage
girl looking, they need to seefemale role models.
If it's any across theinclusion or diversity landscape
, people need to see role modelsthat they go.
You know what.
(33:43):
I identify with them becausethey're similar to me.
I can see, you know, andalthough middle-aged white guys
have had the stage for manyyears, it is time for change.
What are your thoughts aroundthat?
Women in tech what are yourthoughts?
Julie Koesmarno (34:01):
Yeah, that's very interesting.
So here's what I will say, witha little bit of, I guess,
caveat, like I cannot representthe entire women in tech, but I
can share some of myperspectives, which is, I would
probably say, to a certaindegree, unique on its own, but
maybe some of the learnings canbe useful for somebody out there
(34:22):
.
So a bit of background.
I do, or did have the privilegeof having quite strong role
models.
So, for example, like my mom,mom owns a business and it's in
the hardware sort of.
When we talk about hardwareswe're talking about like nuts
and bolts, not software,software, not computer or
(34:42):
anything like that.
So it was pretty uh interestingto see.
So it's sort of like, oh, momcan do that.
So I think I could probablykind of get into more on the
technical side of it as well,like if needed, and there's a
lot of business thinking as wellas math thinking around that.
So that's one aspect of me.
(35:03):
The other aspect of me like youknow how I grew up that I also
consider as privileged too, isthat I did grow up in Indonesia
when I was much younger andstudying computer science was
mandatory.
Everybody got to learn it and Ijust happened to love it.
So I usually kind of like acethe class.
(35:24):
So, coming to a more Westerncountry so as an example,
australia I found myself alittle bit dumbfounded in in
which I chose my elective, mycomputer science, to be my
elective, and I was the onlyfemale.
Also, non-english English is mysecond language too, yes, so I
was like, oh wow, like the worldis very different.
(35:48):
Um, so I had some of the thingsthat kind of helped me think a
little bit more that the worldis just outside of me, that
there are a few different sortof like diverse perspectives.
So what I from personally, frommyself, I try to, I strive to
(36:09):
give different perspectives toall, at least my perspective to
other folks, and also to welcomeothers as well.
Now there are lots of articles,discussions out there where
women are struggling in thisspace.
I'm not discounting that,definitely it's there.
So I wanted to give a bit ofdifferent perspective what it
(36:32):
feels like to be supported, andjust to a bit of a disclaimer,
what it feels like to besupported, and just to a bit of
a disclaimer as well.
It's not like all through mycareer I'm always supported.
That's not the case.
It's just that there are a fewpoints in the career that in my
career that I noticed oh, I knowwhen I think I feel at my best,
where I feel supported, andthat looks something like this
(36:55):
Usually the discussion withinthe team everybody feels like
it's okay to make mistakes.
So we, and it's okay that we areon the basis of learning, we
didn't know, we didn't knowabout something, and then we use
it as a learning experience.
Sometimes it's to do withtechnical stuff like oh, I'm
(37:17):
sorry, I'm kind of new to this.
Can you help me understand X, yand Z?
It could be about co-pilotknowledge or whatever, or what
RAG means or what LLM means, andthen being able to do kind of
like active listening andlearning from there.
But in some cases it might be abehavioral thing.
So if somebody said something alittle bit, maybe a little bit
(37:37):
hurtful, if you can and this isvery, I would say, depending on
the power dynamics and yoursituation right, if you're in a
situation where you can feelsafe then you could say hey, you
know, can we talk a little bit?
I think that the stuff that youmentioned before could be
interpreted differently forother people, so how do we think
(38:01):
about rephrasing that?
Also, what's the intent behindit?
Another thing that I wanted tobring up is just because it's
not your intention.
So if you say hurtful thingsit's not your intention, though,
but it doesn't mean that it'sless hurt for the recipient.
Mark Smith (38:17):
A hundred percent.
Julie Koesmarno (38:18):
Overall overall .
A couple of things that I wouldsay I draw from this experience
is that when I'm in the teamsetting, when it's not zero sum
game, so when we do we playwin-win, I think it's, um, not
just from women perspective, butI think from the team all
(38:40):
around.
We feel like we are respected,um, we can be put into a
position that, hey, I'maccountable for this, my
teammates are, know, making meaccountable for this too, and
you actually, you know, becomeyour better self too.
So those are some of the thingsthat I've learned so far.
(39:01):
Where I started picking, youknow, some of the behavior not
behaviors, but some of thesymptoms I don't know what
characteristics, characteristicsof when I feel a bit more
supported.
Um, then some other experiences.
So, and I have to say, where Iam right now, I happen to get to
(39:22):
witness that a lot.
So very, very thankful of ofthat too yeah, that is so good.
Mark Smith (39:29):
That is so good and I love that story, particularly
starting with your mom andhaving her in business.
That's phenomenal, Julie, it'sbeen awesome to have you on the
show.
Thank you, I look forward tohaving you back again in the
future.
Julie Koesmarno (39:43):
Awesome.
Thank you for having me.
Mark Smith (39:46):
Hey, thanks for listening.
I'm your host, mark Smith,otherwise known as the NZ365 guy
.
Is there a guest you would liketo see on the show from
Microsoft?
Please message me on LinkedInand I'll see what I can do.
Final question for you how willyou create with Copilot today,
ka kite?
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
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.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.