Episode 404: AI Agents and Their Impact on ERP Software and Business

🎙️ In this episode of Dynamics Corner, Kris and Brad converse with recent 🎉 Microsoft MVP Sai Turlapati 🎉. Listen in as Sai shares insights on the evolution of AI, particularly in the enterprise sector, emphasizing Microsoft's significant role in AI adoption. He discusses the importance of prompting in AI interactions and the emerging concept of agents that can automate tasks.
 
🎧 Listen to hear more of the conversation about:
➡️ How prompting is crucial for effective AI interaction
➡️ The practical applications of AI agents in scheduling transformative impact of AI on time management, enterprise applications, and business workflows
➡️ How AI copilots can enhance productivity and efficiency in various industries
➡️ The importance of adapting to new technologies and the potential challenges businesses face in integrating AI solutions.

Send us a text

#MSDyn365BC #BusinessCentral #BC #DynamicsCorner

Follow Kris and Brad for more content:
https://matalino.io/bio
https://bprendergast.bio.link/

Episode Transcript

Speaker 1 (00:00):
Welcome everyone to another episode of Dynamics
Corner.
Copilot, jack of all trades,master of none, but oftentimes
better than a master of one.
I'm your co-host, Chris.

Speaker 2 (00:14):
And this is Brad.
This episode was recorded onJanuary 29th 2025.
Chris, Chris, Chris, that was agood little jingle.
Did you use Copilot to writethat, Chris?

Speaker 1 (00:26):
Chris.

Speaker 2 (00:26):
Chris, that was a good little jingle.
Did you use Copilot to writethat which part?

Speaker 1 (00:28):
No, I did not, I did not use Copilot for that, but
you did say a comment.
You used that term, jack of alltrades, master of none.
And then I realized there'sactually a full quote and this
was very fitting.
When you said that, I was like,ah, very fitting, I like that.
I was like, ah, very fitting, Ilike that.

Speaker 2 (00:46):
And that was fitting, because today we had the
opportunity to dive deeper intothis world of AI, which everyone
seems to be talking about, andthere's a lot of information to
unravel, and there will be a lotof information to unravel in
the future as well, too.
With us today, we had theopportunity to speak with Sai
Charlapati about Copilot, ai andmany other things hey good

(01:16):
morning, good morning.
Hey, good morning.

Speaker 1 (01:18):
How are you?

Speaker 3 (01:19):
doing.
Hey, good morning Chris, Goodmorning Greg, how are you guys?

Speaker 2 (01:23):
Very good, very well, very well, thank you.
Thank you for taking the timeto speak with us.
Been looking forward tospeaking with you.

Speaker 3 (01:32):
Yeah, thanks for inviting me.
I heard a lot of episodes.
I'm really interested to talkto you guys and learn so much
from your podcast, I think today.

Speaker 2 (01:43):
We're interested in speaking with you and learning a
lot from you, or hearing a lotfrom you about some popular
topics that I see a lot ofinformation on and you also
share a lot of information about, which is exciting.
I'm getting old, so it's alldifficult for me and it's very
difficult for me to keep up witheverything that's going on.

(02:04):
It seems that everything'saccelerating quickly and I just
can't keep up, but that's why weget to talk with people such as
yourself to hopefully sharesome insights, to help us get a
handle and a betterunderstanding on some of the
technology that is available tous.
Before we get into the topic,would you mind telling everybody

(02:26):
a little bit about yourself?

Speaker 3 (02:28):
Yeah, sure, my name is Sai Thirulapati.
I am in the IT industry for thepast almost 20 years.
I saw the Y2K.
During the time I was veryyoung, fresh out of college,
trying fresh out of the college,trying to understand that
mainframe transition and otherthings.

(02:49):
Then I saw mobile revolution,then cloud revolution.
So there are these waves oftechnology revolutions that we
saw and I was able to ride thosewaves and recently, for the
past few years, I was veryinterested in the AI space.
So I looked at the different.
Who are the players in the AIspace, especially enterprise AI.

(03:15):
The enterprise AI.
Microsoft, claudie, who is theAnthropic, is the company that
creates this quality, like openAI is having charge upt.
These are the players,especially in the b2c space.
That's how I see it.
In the b2b space, microsoft,amazon, oracle and, you know,

(03:40):
google are the players, butpredominantly I see Google with
Gemini and Microsoft with theirown Azure framework.
They started with Azure withthe backend, trying to talk to
any LLMs, but finally theydecided to just create a wrapper
around it and explore the youknow the LLMs that are being

(04:05):
developed by other players.
So that's how I got interestedin this space and I feel like
the first wave of the impact ofAI is going to be in the
enterprise side, at least on thecustomer service and sales.
That's how I see it, becausethat's where there is a quick
value that enterprises can see.

(04:25):
So in that space I evaluatedwho are the top players in the
CRM and customer service.
Salesforce is one of the topplayers and Microsoft is another
one.
Hubspot is there, sage CRM isthere.
Those are very good players.
So in that I looked at who canreally help the enterprises who

(04:48):
are having the end to end story.
When I looked at it, microsoftis having the teams right
Microsoft teams and Salesforceis having them.
Slack that's the company thatthey bought, so those two are
going to be really competing inthat space for the AI to get the
enterprise adoption.
And one thing that Salesforceis not having especially lacking

(05:12):
is the cloud story, whereasMicrosoft is having the good
cloud story.
I looked at the Google.
Google is, gcp is having cloudstories, so as Amazon, but they
don't have the enterprisesoftware such as Dynamics 365,
erps or CRM, customer serviceand all those things.
Then I felt like, okay, I am inmy 40s, I feel like I need to

(05:38):
bet on one of the real vendorswho are going to take me to next
10 to 20 years.
I looked at Microsoft.
I feel like okay.

Speaker 1 (05:47):
Microsoft is having.
You took your bet on Microsoftversus Google.

Speaker 3 (05:52):
Yes, because Google is not having Chris.
Google is not having any ERP orCRM.
They tried to buy the HubSpotbut they withdrew that bid
recently.
To buy the HubSpot, but theywithdrew that bid recently.
So for any cloud vendors, forthat fact, for any enterprise
companies, to build this CRM andERP systems, it's long, you

(06:15):
know, it takes a long time.
And also, the important thingis the user base.
Right, they can build theefficient software solutions,
but I feel like attracting theusers is a difficult thing, so

(06:46):
you're going with Microsoft forthe.

Speaker 2 (06:46):
B2B enterprise company to have a larger
adoption within the B2B spacebecause of the exposure to
businesses with the existingapplications that they can build
upon utilizing.

Speaker 1 (06:50):
AI.
Well, you covered a lot.
See, it's already a lot onthere.
We're just getting into it, man, I know.

Speaker 2 (06:53):
We're scratching the surface.
We got into it.
I'm still back at see, my mindis still processing.
I'm still back at Y2K, which Iremember when that was the end,
and I almost wonder, you know,maybe would we have been better
off if it didn't then or not?
Uh, but you, you had mentionedmicrosoft.
With ai I mean microsoft, ai one.
To me, artificial intelligenceis a very generic term because

(07:17):
ai encompasses a wide spectrumof topics.
You know, we hear the lg's.
I can't even cover all of thepoints for it because, you know,
a lot of times people justthink of the llms.
You mentioned chat, gpt andrecently we've seen some in the
news some other local largelanguage models allow processing
locally, so it's there.

(07:37):
So, with with microsoft and aiand the adoption, or where you
see the adoption to B2B to adopt, utilize and gain benefit from
the use of AI in theorganizations or increase some
efficiencies, how do you see andposition the Microsoft tools to

(08:00):
be able to use these AIfeatures and what are some
benefits that you see anorganization can get from using
AI?

Speaker 3 (08:10):
Yeah, sure, especially, that's a very
interesting question In the B2Bspace especially, microsoft is
having very good footprint,especially with the Microsoft
365 Office 365 suits and the wayI see is especially the users.
When chart GPT came, that is aaha moment in the artificial

(08:33):
intelligence revolution, right?
They?
Everybody thought that it isgoing to take some time, but the
user interface for chart GPT isa prompt.
I looked at the landscape in theenterprise computing.
Who are having that promptreadily available?
I see there are broadly threeplayers.
One is Microsoft Teams andanother one is Slack, which

(08:58):
Salesforce own, and third one isZoom right Zoom calls.
People are used to this videoNow they started.
You know the charting also.
One is zoom right zoom calls.
People are used to this videonow they are.
They started.
You know the charting also.
So those are the threepredominant players for humans
to have that kind of interactionfrom B to C space where chart
GPT and Tropic, google, geminiand other players are there to

(09:19):
convert that B to C space, thatchart prompt experience, into
the enterprise experience of thebusiness users.
I feel like these threepredominant companies like
Microsoft with Teams, salesforcewith Slack and Zoom, are the

(09:40):
three players that are going tobe really taking this enterprise
AI to the next level.
Those are the three playersthat are going to be really
taking this AI enterprise AI tothe next level.
Those are the user interfaces,because people already have the
experience of using promptingthe chart.

Speaker 2 (09:55):
I hear the word prompting with an AI and I hear
individuals talk about how tobecome a prompt engineer or
prompting tips and tricks forprompting.
What is prompting and how doessomeone come about with the
prompting?
And we're talking with largelanguage models and prompting

(10:18):
how can those be utilized withinthe B2B space?
How does someone understandwhat prompting is and maybe how
to construct a prompt to get theresults that they're looking
for accurately?
But I also want to hopefullyget into also this new thing
that I'm hearing about, which isagents, to where maybe it

(10:39):
expands to a little bit morethan just prompting or typing
for information gettinginformation back, where you have
an agent that can possibly dosomething.
So it will take some tasks thatare possibly repetitive or
tasks that can be automated in asense to allow for someone to

(11:02):
have more time and opportunityto do other tasks.
So how does that all fit withinthe B2B space?
How does the prompting work?
What can you do with theprompting and also then with
these agents that are beingcreated?
I know within Business Centralwe see a lot of news about
Microsoft adding agents andagent previews that are

(11:25):
available and talking about that.
It's not even within BusinessCentral.
I see the word agent everywhere.
I think it's going to be.
I think the word of 2025, if wecould talk about it would be
agentification or agentizingAgentic.

Speaker 1 (11:39):
I hear that too Agentic.

Speaker 3 (11:40):
Yes, yeah, that's a very good question and very
reflective on the introspectivequestion.
What is prompt?
Prompt is nothing, but, atleast in my words, prompt is
nothing but asking a question.
How do you ask a question is aprompt.
How do you ask a question to acomputer?

(12:03):
In this case, the AI bot is aprompt.
The way you ask a question andthe way you respond to a
question is also a veryinteresting leadership insight.
I read a book, or I listened toa book called how Great Leaders
Ask Questions.

(12:23):
So the way we structure thequestion and what is the
strategies that we can use inorder to structure your question
enables the person to gathermore information.
So this prompt is nothing butthe way you ask a question to

(12:44):
the bot or AI agent, right, aion the other side, and the AI
computer or AI bot or AI chart,we call it in the Microsoft
setup, we call copilot, right,ui.
So the way we structure theprompt involves different
strategies.
Right, first, we can give thecontext.

(13:05):
We say that, hey, what is thenews today Is a prompt.
We can ask that as a prompt.
Or what is the news today inthe United States in the
financial sector Is morespecific.
So we are able to structure itand ask a question to get a, you
know, intended answer for us.

(13:26):
So prompt depends on howfine-grained means how specific
we are.
The answer is going to be thatmuch, you know, clear from the
ai agents or ai bots.
So you touch a lot of topics,br.
Brad.
So I agree with you this 2025is going to be the age of agents

(13:47):
.
You know, when we talk aboutagents, I remember the movie
that I watched in 1998, matrix,right, I'm sure everyone
remembers about that movie.
You know the agents.
So the difference between theway I look at it is the
difference between agent is anautonomous thing.

(14:08):
That's what Salesforce is alsocalling them.
And Salesforce came up with theagent force as one of their
solutions and they are goingfull-fledged.
How Microsoft came with Copilot,satya Nadella, who is the CEO
of Microsoft, very clearlyarticulated that Copilot is the
user interface that humans aregoing to interact with the LLMs

(14:33):
or the AI machines, right, andthe backend is going to be the
agents who are going to do thework.
If we try to do that in the,you know, correlate that space
into the Power Platform, I feellike agents are nothing but

(14:54):
Power.
Automate, right, they arenothing but a Power Automate
workflows right.
Co-pilot, when the userinterface, when user prompts or
ask a question.
That goes to the agents,microsoft.
Interestingly, in one of theirdocumentation they referred
agents in three ways One is aresponsive agent, another one is

(15:17):
a task-based agent and thirdone is autonomous agent.
So I feel like, chris, you areknow when we talk about agent,
which can go and do a task andcome back, it is like a power
automate flow, right?
People who are aware of thisMicrosoft power platform knows
what power automate is, which isnothing but a RPA space.

(15:39):
Uipath is another company thatyou know they do in the RPA
space that they provide thesolutions.
So for our context, agent isnothing but a you know, a
software program in the back endthat goes and completes a task
without giving the information.

(16:01):
Then what is the differencebetween?
Now comes the question what isthe difference between power
automate and the agent?
Right?
Power automate we use to go inthe power automate.
If I want to create a flow, Ineed to go and drag and drag and
drop all the requiredcomponents.

(16:21):
What is the trigger, what itneeds to do?
The power automate means send ita email.
Let us take a simple use case,right, if we want to read an
email based on the incomingemail, I just want to create an
Excel sheet or Word document andsend that information back to a

(16:42):
team.
If I take that use case inorder, for If I take that use
case, in order for us to do thatuse case right now in the Power
Automate, I need to go andcreate a trigger, say that, hey,
incoming email is the triggerto this email box.
Once we get that email, then dothis processing, read the email
and create that Excel or Wordand send that information to the

(17:04):
teams.
I need to go and create thatExcel or Word and send that
information to the teams that Ineed to go and do that.
But Microsoft, now recentlythey created a copilot for Power
Automate.
Now I can go to the copilot andsay that, hey, create this
workflow.
So this workflow of reading the, you know, anticipating for the
email and reading the email andcreating a Word document or

(17:27):
Excel and sending it to them.
So, stepping back really quickSai.

Speaker 1 (17:32):
You mentioned Copilot , basically more of a.
The way I look at it soundslike to me Copilot is more of a
translator.
You ask a prompt of what youwant based on what's available
for you within your maybe tenant, then it chooses the correct
agent to respond.

(17:52):
So it's almost like atranslator.
Right for that prompt Because,as you know, in the B2B space,
when you're creating or you'reinteracting with Copilot within
your organization, it shouldonly respond based upon what's
available to it.

Speaker 3 (18:09):
Right.
So you are right, chris.
So the Copilot is like you said.
It's basically an interface.
It does some operation, itmanages the agents.
You can say that it's anorchestration piece where it
takes the information from theuser and, based on the available
agents, it will direct theagents, orchestrate the agents

(18:33):
to go sequence of tasks and comeback and provide the answer to
the user, to the human or to theuser in the NLP, natural
language processing.
So now the way we interact withthe co-pilots or AI agents is
completely changed.
From the mouse, we take thatand click that different buttons

(18:54):
to get the information.
Now we are using naturallanguage processing to talk to,
like how we are able to talk toother human, like how we are
discussing, we are able to justenter the information to the
co-pilot.
Microsoft is having their ownlab called co-pilot.
You know Microsoft Labs.
They are experimenting withvoice also.

(19:17):
So, like how we are discussing,they have a co-pilot voice, the
co-pilot voice.
We can just enable the voiceand we can say that, hey, this
is the task that we want to do.
Then you know, it can go aheadand create the agents and
orchestrate the agents and comeback and with the answer.
So in one of the recentinterviews also, I think, satya

(19:41):
Nadella, ceo of Microsoft, hetold that SaaS kind of you know,
in the future SaaS may be.
What is SaaS applications?
Saas is software as a serviceapplications.
Right, they are basically aCRUD.
Applications means they arehaving a database On the top of
the database.
The user interface provides theuser to interact to perform the

(20:06):
CRUD operations.
If we take CRM right, crm ishaving a sales module in that
there are certain databasetables which are in the Power
Platform called Dataverse.
Sales module provides the userinterface for the users to go
ahead and create codes, purchaseorders, leads and opportunities

(20:34):
, all those things.
In the future, what is going tohappen is people are expecting
that may be sooner, maybe withinthe next few years.
Instead of user going to thesales application and entering
the information, people will goto the prompt copilot, sales
copilot and they say that, hey,this is a new lead that I got,
this is the you know.

(20:55):
Take the picture.
Say that, hey, create a leadinformation in the sales of
Dynamics 365.
It should be able to createthat information.
So the user experience itselfmay be, you know, completely
changing the way users interactwith these enterprise
applications.
Maybe really changing.

Speaker 2 (21:17):
That could take me down a completely separate path
because Chris and I recentlyspoke about that as well as far
as how we interact with data,how we retrieve data and having
the ability to use naturallanguage to interface with that.
But I'm still trying to go wayback to the beginning of

(21:40):
prompting to get information out.
How do we come up with a andhow do we learn and how do we
know to come up with the properprompt either for to go back to
the points that you hadmentioned, either it's data
retrieval or language andlearning I type, you know,
create me a picture or ask someinformation based upon the data
that the model has been trainedon or in the construct of what

(22:05):
you and Chris had mentioned,with the Power Platform to
utilize Copilot Studio in asense, which I want to get into
to create these tools for us,basically our own agents.
But where I get confused is wementioned task-based agents how
is their variability in tasks?

(22:26):
Because I still say, somethingthat I tried to do, I wish I
could do, is even something assimple as scheduling, taking my
emails, taking a look at mycalendars to be able to
automatically reply, like evenwith the podcast, for example,
we do a lot of scheduling of theguests, such as yourself with
the podcast, with the.
We do a lot of scheduling of theguests, such as yourself with
the podcast, with thepre-podcast planning calls.

(22:50):
Chris, you have to fix thatPre-podcast planning calls to
the actual schedule of therecording taking a look at
calendars, taking a look at timezones to offer and suggest
times that best fit based uponavailability time zone and such
times that best fit based uponavailability, time zone and such
.
There's a lot to that, and isthat something that could be
done and how could you do that?

(23:10):
Is that multiple agents withinPower Automate?
No-transcript.

Speaker 1 (23:42):
Right, so now you give a it's all within that
space.

Speaker 2 (23:46):
So, utilizing that, how could I do that?
I hear a lot about Copilot andI hear a lot of things that we
have agents that can do anything.
I'm just trying to see apractical use and example of it.

Speaker 3 (23:57):
Yeah sure, so let's take that use case that you
mentioned about the podcastright For us to create this
Microsoft AI agents or Microsoftco-pilots broadly.
There are two ways that we cando it right now in the Microsoft
platform.
One is using the.
Microsoft came up with aco-pilot studio that is part of

(24:19):
the Power Platform that providesthe tools and knowledge bases
and inbuilt agents also thatenable the user.
That's a low-code, no-codeplatform Copilot Studio, where
the users can go ahead andcreate the AI agents.
And another way to do that inthe Microsoft platform is Azure
AI Foundry.

(24:41):
Microsoft just recentlylaunched Azure AI Foundry.
Microsoft just recentlylaunched Azure AI Foundry, which
is based on them.
We can go ahead.
We can use the Azure AI Foundryand create the agents using
different LLMs that areavailable, such as we can use
OpenAI Microsoft is having 49%stake in the OpenAI, so they

(25:04):
create exclusive access to theOpenAI models or we can use
Anthropic models, or we can useLAMA, which is Meta's open
source AI models.
So broadly, we can do it in twoways.
One is Copilot Studio MicrosoftCopilot Studio or Microsoft
Azure AI Foundry.
For our conversation.

(25:26):
I have good experience in.
You know, I created a couple ofagents in the using Copilot
Studio so we can do the use casethat you mentioned using the
Copilot Studio.
Copilot Studio is a very easyway for us to create the agents.
Previously Microsoft used tocall as co-pilots.
They renamed it a few monthsback to agents.

(25:47):
So to create any agent we needbroadly two or three things.
First one is what is aknowledge base right, based on
what the agent need to createthe information.
Second one is tasks.
These tasks are nothing but thetasks that just you outlined to

(26:08):
create this podcast.
We need to look at, you knowfirst, evaluate this.
You know participants, sendthem an email and have the
review session so we can create.
We need to break down intodifferent tasks and that task.
Microsoft is very good,especially in the Coopilot
Studio.
They came up with a lot ofconnectors.

(26:29):
We can connect with the Outlook, which is a native thing, so we
can easily connect to theMicrosoft Outlook and create a
task and send an email If wewant to talk to any other
databases or like Riverside ifwe are trying to use, task and
send an email If we want to talkto any other databases or like
Riverside, if we are trying touse.
We will go ahead and seewhether the Co-Pilot Studio is

(26:50):
having any Riverside connectors.
If not, users can create thatcustom connectors.
So Microsoft enabled all thesefeatures for the low code.
No code developers to createthis kind of connectors to
create the agent.
So the agent can be created.
First, this podcasting use casewe can break down into tasks

(27:12):
and each task we can go aheadand create.
It is similar to creating aworkflow in the Power Automate.
You're muted, by the way, brad.
You are muted.

Speaker 2 (27:22):
Brad, don't tell anybody, because I was getting
excited and I had to mute myselfbecause I wanted to hold back.
So I can create an agent thatwill send an email to Sai.
Sai, we'd like to speak to youon the podcast, are you okay?
I'm simplifying.
It may be one use case In theother cases where individuals

(27:45):
contact us and say they wouldlike to speak with us about a
topic which we enjoy gettingthose emails as well.
So I can say email Sai, ask himto be on the podcast.
You'll reply yes.
So the agent can reply to thatemail, knowing that the original
email was sent out as a requestto the podcast.

(28:06):
Response is whatever verbiageis yes, and then the agent can
respond and say okay, let's dothis, let's set up a planning
call.
Is this time good for you?
Or here's these times that aregood for you?
Based on rules or based onBased?

Speaker 1 (28:23):
on our calendars.
If we we could do this, thiswould be amazing.

Speaker 2 (28:27):
So I can send out the email.
The response would come back.
It would automatically senddates based upon a calendar for
availability.
The participant would be ableto then respond with this works
best for me and then it wouldautomatically schedule and put
in the text that I like to use.
And all that and the link, yeah, because the studio link is

(28:50):
yeah, that's, that's correct,right.

Speaker 3 (28:52):
Yeah, that's correct, chris.
So what you know, now we aregetting into very we need to do
this afterwards.

Speaker 2 (28:57):
If we can really do this, I would like to schedule
time.
You can certainly set it up isthis something you can do and
then send to me?
Yes, I think we can try that.
We'll do a follow-up because Iwant to see this in action,
because that is such a goodexperience and good use of the

(29:18):
tools, as well as saving timeyeah, not only that, you that
you know you will have, you knowthis Chris and Brad will have
their own agents right To sendthe email to schedule this.

Speaker 3 (29:31):
I will have my own agent.
So as soon as I, that is whocould do my email right.
So as soon as I see any requestor any information if I want,
my agent will respond to youragent.
I can create a small agent, saythat, hey, if I am going for
this podcast or speaking once Iget an email, I create a small

(29:52):
agent and say that, hey, justrespond back with my email ID,
with my calendar availabilitynext couple of weeks to your
agent.
So this is going to be reallyagent orchestration.
Right At your end you will havea couple of agents which will be

(30:13):
triggering the email orreceiving the email.
I can just next time I willsend an email say that, hey, I'm
interested in your podcast.
Or I will just ask co-pilot saythat, hey, next couple of weeks
I am interested to.
I have this time I want toreally talk and get to know more

(30:33):
about what is happening in thedynamics with Brad and Chris.
So I will just ask the co-pilotand the co-pilot goes and talks
to my agent and send an emailto you guys and your agent can
pick it up and look at youravailability and schedule and
confirm something back to myagent.

Speaker 1 (30:53):
There's actually two places where I can see this
working Right now.
Our website has a place for youto be a guest, right, you fill
out your information and stuffand we get an email.
So we could use a PowerAutomate to collect that
information and, based on thatinformation, then we can use
Copilot to act upon that.
Where it looks at our calendars, make sure that it answered you

(31:17):
know we got all the informationit needs and then it can, you
know, schedule that for us.
And number two we can even putcopilot on our website, probably
, and sai can interact with it.
It's going to ask all thequestions that the form would
have asked anyway, collect thatinformation based on size
responses, notifies us and looksat our calendars and let us

(31:40):
know like hey, si's interestedin this and then schedule it out
and we just show up and have aconversation.

Speaker 2 (31:46):
I I like the use case because it's to go back to
where I started with.
This is.
We hear all this ai agent, wehear prompting, we hear models,
we hear all this, but I hear it.
Well, you can do anything oryou can do specific tasks.
Now I'm trying to just put myhead around something as simple

(32:08):
as scheduling with somebody tobe on a podcast.
Somebody could do this on theirown.
For something else, even lawnmaintenance, if somebody has a
landscaping or a lawnmaintenance company or even
electrical services, you knowany type of scheduling that you
need to go through.
We usually have an individualgoing back and forth.
I know often I'll use what theyused to call fine time or

(32:31):
whatever.
That is where you can say, okay, here's several dates and times
, pick the ones that work thebest and then from that.
But that's, to me, is not suchan elegant experience sometimes.
So I would like the morepersonal interaction of I'm
sending an email to Cy.
Step one, chris.
I like the idea of also takingingesting on the inside.
Then all of a sudden stuffshows up on the calendar and

(32:53):
then we just just do it.
I think that's also great.

Speaker 1 (32:56):
I think you can even prompt it where like, do not
schedule anything just withinthe space because maybe you're
already busy, and then look at,you know if there's already
existing one and you can specifyit's the only schedule between
these days or between thesetimes.

Speaker 2 (33:12):
Oh, we have to do this.
Okay, so we can set it up towhere we have an agent in Power
Automate that will send an email.
It will send an email based onme just saying send an email to
Sai to be on the podcast, right,yeah.

Speaker 1 (33:30):
Just like that.

Speaker 2 (33:30):
We can easily do that , yes, and then we'll send an
email to Sai with a templatewith the information that's
pertinent to Sai so that heunderstands what the podcast is.
Then you'll reply.
The agent will read the replyand we can say go look at this
calendar and this calendar andpropose some times based on size

(33:53):
, time zone.
I like this.
That is important.
I'm trying to think of thevariability here because these
are the scheduling challenges.
We recorded all hours of theday to make accommodations are
the scheduling challenges.
You know we recorded all hoursof the day to make
accommodations for everyone'sschedule.
That that's going to so howmuch how much time you spend,
brad.

Speaker 3 (34:10):
I think uh looks like it's a lot of efforts for you.
You know, to host this kind ofpodcast right.
Look at the look at the justcalendar.

Speaker 2 (34:17):
Simple task of calendar it's looking at
calendars, going back, makingsure that it's a lot of
appropriate for the guestsbecause, as we say, it's it's
anybody who's been on.
We, you know, in the planningcall, we talk about that when
they're at their best, when itfits them.
We try to work around trips ifsomebody has conferences to go
to, for example, or if there areholidays and and sometimes

(34:41):
individuals don't mind, you knowwe'll do a recording on a
holiday or something.
So there are some variables inthere that, based upon, we may
need to find out which timeswork best for you, but it is a
lot to juggle schedules for manycalendars.

Speaker 3 (34:55):
So how do, how do you guys do?
Do you guys outsource thatpiece of?

Speaker 2 (34:59):
work we do.

Speaker 3 (34:59):
outsource it to me piece of work to me.

Speaker 1 (35:09):
Yeah, man it's a lot of work, right then a lot of
energy, guys.

Speaker 2 (35:10):
The scheduling is a lot of work.
The scheduling is uh, I'd wantto say it's a full-time job, but
to try to do proper schedulingand I like to do things properly
as well as you see, when you gothrough the experience, to make
sure that everybody has anenjoyable experience as they go
through this with a little bitof a personal touch as well.
But, it does say take some timebecause we have several calls
per week that we do on top ofeverything else.

Speaker 1 (35:32):
You know, it'd be fascinating though, because once
we get all this solved right,it's to have a system, or maybe
co-pilot or an agent, where hetakes all of our files and says
I want you to edit this for ourpodcast video and it just does
it for us.

Speaker 2 (35:51):
Right now it's all manual man, we do it man we do
it well, chris, does the thepost production you know we have
a process and it works well.
Because of timing, I'll do alot of the scheduling,
interfacing with the guests thatcome on that we are extremely
appreciative of everybody thatspends the time with us.
Time is extremely valuable.
I know to me personally andChris and I talk about it

(36:12):
because it's what we have.
Once you use it, you don't getit back.
You don't get a redo of aminute.

Speaker 3 (36:19):
You are a wise man, Brad.
People realize that very late.
I am realizing now.
I think time is the real.

Speaker 2 (36:36):
It is.
It takes you.
You do have to get to a certainpoint in life, I think, where
just maturation in life, thatthis is something I wish.
Everyone always asks me whatwould you tell your younger self
?
That's one of the things Iwould tell myself is one listen
to those that have gone througha lot because their experiences.
I'm not saying you have tolisten to the experience as far
as following, but sometimes justlistening objectively to
somebody's experiences and maybelearn from them instead of
thinking, ah, they don'tunderstand, I can do this, I can

(36:57):
do that.
And also time, you know whereyou spend your time and what you
spend your time on.
You cannot value anything morethan that.
But to go back to what wetalked about, so I do a lot of
the scheduling, chris does thepost-production.
Hopefully at some point we canincorporate some ai into it, um,
which thankfully riverside hasadded quite a bit that we can do
some stuff.

(37:17):
But uh, chris, I don't want tojump the gun yet.
I'm still trying to go back tothe scheduling to save myself
some time and we'll set this upand I won't even tell Chris and
I'm like I have to go throughall this scheduling.

Speaker 1 (37:26):
I think, like you said, you put it well perfectly
about the time spent, where Ithink co-pilot AI in general is
really, really important for thefuture, because at that point,
as we use it more, we don'ttrade time for money.
We get to a point where I wantto trade time for experiences,

(37:48):
right, and and that's going tobe, um, that's going to be
important in the future for it,for my, for my view, because, as
like you said, sai, you know,you want to make sure you're
you're utilizing your time inthe right places and not trading
it for money every single time.

Speaker 3 (38:07):
Life is short.

Speaker 2 (38:10):
It really is.
It's um.
You know, you go through phases.

Speaker 1 (38:14):
I thought we were now getting philosophical and I
don't mean to digress, maybeI'll just stop.

Speaker 2 (38:17):
I could get on the philosophical road forever but
um time.
So we'll go back to theefficiency of time.
So we can set up an agent tosend an email.
We can set up an agent to readan email and then, based on the
contents of the email, work onscheduling the podcast.
I have to see this work.

Speaker 3 (38:36):
Yeah, that's a good use case, at least in 2025.
Maybe next, 2026, 2027, we justneed to talk to that.
You know co-pilot or the UI,the prompt say that, hey, this
is what we want to do.
It has to.
It may be going and doing allthis.
You know scheduling andcreating all this for us.
Maybe right now we need toenter the prompt on the keyboard

(38:59):
In the future.
You know there are NVIDIA.
All these companies areinvesting heavily on voice-based
, so we just talk to them.
Hey, do you guys?
I have a question Do you guysrecently OpenAI released
Operator?
Did you guys look at thatOperator?

Speaker 2 (39:17):
demo.
I watched the demo of again youneed to have from what I read,
you need to have OpenAI Plus orwhatever that means, Right, $200
.
The expensive plan, which Iunderstand, but that was
impressive as well.
It would go to a website andinteract with the website.
I saw it do the scheduling, Isaw it do reservations and a

(39:39):
number of other things.
That is impressive.
That's interesting.

Speaker 3 (39:42):
That is impressive.
Yeah, interesting, that isimpressive.
Yeah, I have a question forboth of you.
Given you guys have so much ofexperience in the IT side, right
?
So, technology-wise, in theleadership roles and all those,
how do you see this enterpriseAI evolving?
So I shared my thoughts, right,and going back to that
mainframe era.
I know a lot of companies likebanks, insurance companies.

(40:03):
Going back to that mainframeera.
I know a lot of companies likebanks, insurance companies.
They are still on mainframe,given the architectures you know
in the enterprise computing.
You know it is very difficultto move to the latest and
greatest, given SLAs and youknow lawsuits and all those
things right.
So on the Java side, I worked inSun Microsystems in India for

(40:26):
almost three years, the guy whoinvented Java also I was able to
.
I was part of one leadershipcommittee way back, so I was
able to meet that guy, jamesGosling, who invented Java and
it's a very interestingexperience.
But when I put that also in thecontext, java, the latest
version is Java 20 plus rightNow, oracle, bot, sun, they have

(40:49):
20 plus, but in production theywill have old versions of Java.
Still, I know Java 8, which wasreleased way back in 2006-2007,
their enterprises are using it.
If we put mainframes and Javaversions in the enterprise
penetration right In theproduction environments, people

(41:09):
will be doing a lot of testingand POCs and sample projects.
But for enterprise applicationsor CRM kind of applications,
how do you guys see this AIgoing to, the transition or
adoption of this AI in this ERPand CRM space?

Speaker 2 (41:29):
That is a challenging question.
In my opinion, you brought upsome key points.
It's where is an organizationin its journey?
Which systems do you have?
Also, which systems and whichtools are available?
I think you need to have acombination.

(41:50):
Should you always have thelatest and greatest?
I don't know.
I think, in my opinion,sometimes there's risk and you
need to evaluate what you use.
I think for those more matureorganizations, there may be
pieces that you can plug in.
So if you think of even goingthrough to bring it back to a
point, to a business, centralimplementation or implementation
of an ERP application, it's amatter of architecting a

(42:14):
solution with the right piecesand putting those right pieces
together to get the desiredresults.
But individuals also need toalmost change their way of
thinking at some points to askquestions outside of the
predefined constraints, Becausea lot of times people make
decisions based on the past.

(42:36):
I know, running through thiswoods I ran into a bear.
So now I'm going to make sureevery time I come through this
point I'm going to ran into abear.
So now I'm going to make sureevery time I come through this
point I'm going to run into abear.
But now I need to make sure, ifI go down this path, I do all
this stuff but in reality, thatbear may never be there again.
So you have to take off theconstraints of the limitations

(42:58):
you had based upon the past andnot think that you need to do
something in totality to whereyou have such a radical shift,
but maybe compartmentalize thepieces to incorporate those
changes Again the efficiencythat you can get in the AI to
where you can increase theadoption within your
organization and ensure thatit's going to give you the

(43:19):
desired results as well.
It's a lot there of what I amsaying, but I think
organizations need to evaluatewhere they can get the benefit
of using a tool, making surethey use the right tool and
don't use the tool just to usethe tool.

Speaker 1 (43:38):
Yeah, that's a good point, Brad.
From my perspective, there aretwo different paths, because
there's still some human element.
As you know, when someone usesa technology on their personal
life, they typically bring thatto work, expecting to do the
same thing.
There's still a lot ofeducation that still needs to

(44:03):
happen when it comes to AI.
I've had a lot of conversation,a lot of people, even
day-to-day people, that talksabout AI and they think it's a
one one solution fits all and,as we know, in our space, that's
not always the case.
That's not the case at all,because now we're talking about
agentic or agents that are usedin the back end to do some

(44:28):
specific task or give youresults or just have a plain
conversation of when you'rehaving that utilizing ai.
So I I still think we have alittle bit of time that we need
to educate everybody that thereare differences between the two,
because you talked about Geminiright as another LLM, but then

(44:50):
you also have Copilot and sofrom a public perspective, it's
just an AI to have a goodnatural language conversation,
but in our space, in theenterprise space, that's not the
case.

(45:18):
But in our space, in theenterprise space, that's not.
We're going to have systemstalk to each other and what
comes to that means there aregoing to be things or,
unfortunately, positions thatare going to be replaced because
of those co-pilots or becauseof those agents, and I think,
from a business standpoint,that's going to be a place where

(45:42):
you need to plan.
You know what does that?
What does that mean for yourbusiness?
Um, that means more time foryou, maybe more time to be more
creative, but even then,creativity could be replaced as
well, it's like ai, it's, it'sone of those never-ending cycles
, but we also have to.

Speaker 2 (46:00):
It goes back with time.
It's sometimes you have toworry about what's in front of
you versus so far into thefuture, because you don't know
and nobody can predict what willhappen.
Uh, the any businesses haveevolved since back in the early
days when they use ledgers withink pen, inkwells and pens,

(46:20):
right quill pens, and now thenyou had the ballpoint pen and
the pencil, and then you went tocomputers with spreadsheets.
So there's always been anevolution in of efficiencies and
gaining of those efficienciesand then just a reallocation of
talent to do those tasks thathaven't been to the point where
they have been as made, optimal,I guess you could say, or added

(46:43):
the efficiencies.
So it's.

Speaker 3 (46:46):
Yeah, I think both your points are very valid and
very interesting perspectives.
The way I look at is the ROIright.
As a business, they will seewhat is the ROI on their
investments.
So, especially with the teamssummarization now when I am

(47:06):
using teams in my workplaceafter the meeting previously I
used to take the notes and allthat information.
There will be someone who takesthe notes and share that after
the meeting.
Now, with the team scope, wecan get a summary of the meeting
pretty quickly.
Perfect, that is a very goodROI.

(47:27):
For me, that is a great usecase.

Speaker 2 (47:29):
I don't mean to cut you off, but with Teams turning
on the transcription to recordit.
some individuals get a littlenervous, thinking I'm going to
do the video recording orsomeone has the recording, but I
agree with you.
Recording, or someone has therecording, but I agree with you.
Just something as simple asdoing the transcription of the
voices gives you the benefit inthe meetings that those
participants can pay attentionto what's being discussed

(47:51):
instead of worrying about thenotes that they have to take,
because you cannot do both tasksat once.
If you're spending time tryingto write the proper notes,
you're not listening.
I don't care what anybody says.
They think that they canmultitask listen and basically
listen and talk at the same time.

(48:12):
So I just want to bring upsomething as simple as that is a
huge gain.
And you get actual summaries,and I don't know why.
I wish I could have it setwhere it says okay, record any
call that I jump into for thetranscription so I don't forget,
because there have been timeslike, ah, I wish I had this on

(48:33):
and I forgot.

Speaker 1 (48:34):
Do you remember, on meetings like that, where you
have someone's responsibilityand that's all they did was
note-taking?
If you guys recall back in theday, where you sit in a
Conference you have somebody sitin the back corner, that's all
they did was take notes, right,that's my point.
Right, like there's gonna be ashift where that role is no

(48:54):
longer needed in.
It's a lot more accurate fornote-taking, right, and and so
it's.
It's always always listeningwhen a human may be distracted,
and so it's always alwayslistening where a human may be
distracted.
And they forgot a specific note.
Now, with Copilot, to be ableto do that for you and summarize
and even ask, like, what wasthe action items out of this
meeting?
Because you want it to be aproductive meeting, it's going

(49:17):
to tell you, versus having tolike ask that person, say, hey,
can you type that all up, andthen by the time you get those
notes, it's the end of the daywhere Copilot can give you that
information right after thatmeeting.

Speaker 3 (49:32):
Yes, I think that you are right that jobs and all
those they need to be retrainedor they will be going into a
different place in this era.
That is one of the reasons inthe Copilot Studio.
Microsoft enables us as soon aswe create the agent.
It allows us to publish to theteams first Teams are.

(49:53):
It gives a different channelswhere we can publish.
It Looks like the teams is oneof the very efficient way to
interact.
And another use case alsorecently I was working with one
of the very efficient way tointeract.
And another use case alsorecently you know I was working
with one of the clients.
They had a lot of knowledgebase.
They went with a bigimplementation right of their
ERP but that was not successful.

(50:15):
So they were trying to evaluateand we got RFP.
I was trying to me and my teammembers were trying to look at
their business models orbusiness rules.
What exactly is their business?
So I cannot share the clientdetails, but it is in the
healthcare sector.
There is a lot of information.

(50:35):
They went for the ERPimplementation for almost one,
two years.
It was not successful.
We got a RFP.
So generally the process was gothrough all the documentation to
understand their business,specific business, because you
know their business process andwhat are the challenges that
they faced.
So what we did was we tooktheir Word documents, their PDFs

(50:59):
, their audio, their videorecordings and downloaded the
transcripts.
We combined all those thingsand we uploaded to the
SharePoint and created aco-pilot on top of it and
started asking questions to theco-pilot Say that hey, what is
the business process?
Can you explain in broad I knowhigh level 10 steps.

(51:20):
Surprisingly, it was able togive us at least six, seven
steps correctly.
So we started fine tuning theprompts and adjusting, given
what happens is this copilot,once it goes to the knowledge
source, in our case theSharePoint documentation, the
backend, it indexes and itcreates a structure so that the

(51:43):
co-pilot can efficiently go andread and give the response.
So we need to do some kind offine tuning.
But instead of going throughall the documentation, videos,
transcripts, I asked what arethe different statuses, what are
the major modules?
It was able to give me all thatinformation.
That was really, you know,reduced my work at least 200

(52:08):
hours just to go through thedocumentation.

Speaker 2 (52:10):
It's incredible the use.
I just get excited and I can gooff on tangents because think,
now, having all that informationreadily available, you also
don't need to memorize and ifyou can retrieve the information
quickly, now you can prompt toget the information back without
having to spend time searchingor memorizing or going through.

(52:33):
Nothing's perfect, but even ifit can get you 80% of the way
there, or even something assimple as you had mentioned,
that it can do surprisingly wellmost of the steps or most of
the things that are outlined, atleast it gets you started and
it can show you the referencedocuments where you can look
more.
I'm waiting for the day where Ijust it's like we plug in the

(52:54):
microphone jack into a computerto speak that you have a little
jack you plug into your head andyou just think and all of this
information will come into yourbrain and you know.
It's like having an externalhard drive Exactly, you know.

Speaker 1 (53:07):
What's interesting is that you gave this a perfect
example on a specific industrythat's.
You know, even in legal it'sthe same thing If you ever dealt
with.
There's a discovery and there'slike tons and tons of documents
and typically a paralegal wouldbe the one that's doing that.
Right, they would have toreference a specific document,

(53:28):
and it's a lot of writing andall this stuff.
Now you can just use Copilot todo that and be able to
summarize or even search for aspecific thing like, hey, did
this person say this?
And then they'll say yes, theydid for that particular topic,
and then it's going to referencewhere it found that on the
documentation.
So that's actually a perfectuse case.

(53:51):
Sorry, and it goes back to theco-pilot that we were talking
about, where a co-pilot issomeone that can when I say
someone, see, I'm referring it,now, it's like a human being
someone that can, when I saysomeone, see, I'm referring it,
now, it's like a human being,it's a tool that coordinates all
the other agents.
Another perfect use case forthat would be if a client

(54:12):
interacts with your co-pilot andsays, hey, I want to look at
your products that are availablebased on my description.
I'm asking.
Then it could look at your datadatabase and looks at hey,
these are the items availablefor this business and then if
the client's interested inpurchasing that, then that same

(54:33):
co-pilot can call another agentto create a sales order.
So you got to look at that way.
When we're talking aboutco-pilots, that's another use
case.
When it's calling multipleagents based on different tasks
One's informational and then theother one could be a task where
it creates a sales order andthen submits it to maybe a

(54:55):
business central.
So many different ways, usecases, so many different ways.

Speaker 3 (55:00):
Use cases.
Yeah, that legal area is legalvertical, especially legal
industry, like healthcare isalso.
So I was very interesting thisAI is going to disrupt more.
I was reading somewhere aboutwhich area AI is going to have a
very quick ROI.
Looks like it's interestingly afinancial sector, because

(55:24):
everything is really predictable, mathematical.
So what they are saying is inthe financial sector AI will
have a lot of impact, means itcan bring a lot of ROI In terms
of healthcare.
It can unlock a lot of newmedicines and solve and come up

(55:45):
with a lot of breakthroughs inthe healthcare sector.
So, as in legal, which is moreon the documentation side.
You are right, Chris, theparalegal work, like meeting
notes, paralegal work also couldbe now backside.
You know, take a back step, orparalegals can use the copilot
to come up with theirunderstanding and revalidate.

Speaker 1 (56:08):
Finance is interesting, though, because
it's a large, natural, largelanguage model, but it doesn't
does it do a good job crunchingnumbers, though, if it gives you
a bunch of data.

Speaker 2 (56:21):
I think you need to have the right tool for the job.
That this is where it goes backto.
If you're trying to use ahammer to do math, it's not
going to work.
If you're going to use acalculator to do math, it's
going to work.
And this is where I thinkhumans have this natural ability
to want to destroy.
You build a robot.
People will start throwingthings at it.

(56:42):
Why?
Because Because it's a robot,and I think the same thing.
When this first came out,everyone's like oh, I trained it
to do four plus four equalsnine and not eight.
It's silly, but you have tostep back.
That's not what it's supposedto be doing.
It's supposed to retrieve andsummarize and show information,
not do math.
So even go to the agent.
The agent that sends the emailisn't going to be the same agent

(57:06):
that responds to the email inour scenario.
So it's a matter of using theright tool.
You mentioned model Sai or theright agent for the job.
So I think, with finance,depending upon listen, finance,
if you're talking investmentsand finance that whole market
can spiral if you just let it go, ai out of control, because it
will just read the patterns andrespond to the patterns and you

(57:27):
could just have a sharp crash ora sharp spike.
But I think, from anorganizational point of view,
you can use AI to help withfinancial information, financial
reports or even some analysisof information, which you need
the right agent and you alsoneed to have the proper data.
This topic can get me all of itwhen it comes to AI within the

(57:52):
world.
It's great, but to go back tonow, business Central, to go
back to Copilot, agent PowerPlatform and the use of AI
within that space is where Ireally like to focus, because I
think, brad, the enterprise,especially enterprise, ai,
Microsoft, the way I assessed acouple of years back, I think

(58:14):
that still holds good.

Speaker 3 (58:16):
The feature also looks like Microsoft is going to
win a lot of Dynamics 365, er,pcrm and go head-to-head with
Salesforce, hubspot, oracle allthe spaces given their ecosystem
, microsoft ecosystem,especially with Cloud, azure,
with data.
They streamlined their dataplatform.

(58:38):
Now they are calling it as aMicrosoft fabric.
So that is a.
Satya told that that is thegreatest enhancement that they
did to their SQL Server dataplatform.
Right, sql Server, power BI.
They have ETLs I work withtheir ETLs like Azure Data
Factory and SSIS, ssrs, allthose things you know.

(59:00):
They all club together and nowthey are calling as a Microsoft
Fabric.
Data is one of the veryimportant things for these
co-pilots or agents to work.
So Microsoft really, you know,nailed it.
It is competing.
I have some experience withSnowflake also, which can go to
different clouds and gather theinformation.

(59:20):
That's what.
I was a data architect for acouple of years.
In 2017 to 2019, I worked forFord and FedEx as a solution
architect.
I started with the programminglanguage.
Then, in one of the meetings,one of the project managers said

(59:41):
, hey, that is a matrixorganization.
I asked what is it?
Project managers said hey, thatis a matrix organization.
I asked what is it?
He said, hey, that is a PMPthing.
So I said I'm interested tolearn.
Can you help me?
He said go and get PMP.
So I went and I got PMP and inalmost 2009,.
They gave me a big project.
I ran the project.
I understand how to run theprojects and what it is.

(01:00:01):
I got real good respect for theproject managers.
Then I looked at the space andarchitecture is the technology
thing that I like.
So I worked as an architect.
I went and I got my TOGAFEnterprise Architecture
Certification to look at thespace, this IT, differently.
So that gave me a differentlens when I look at all these

(01:00:25):
things and put this AI journey.
Microsoft is having MicrosoftFabric ecosystem, which is a
data which is core to thisco-pilot.
It's a knowledge base, and theystreamlined their security
landscape.
They have Azure or Azure KD andall those things.
Now they are calling it asAzure Entra.
So Copilot is there.

(01:00:46):
Now if people have Copilotenabled in their Azure security,
they can go on just prompt itand say that, hey, what are the
security risks?
There are a lot of tools, fromMicrosoft to Microsoft Defender
and all those things.
So Microsoft platform, you know, especially the implementers or
partners.
There are a lot of tools fromMicrosoft to Microsoft Defender
and all those things.
So Microsoft platform, you know, especially the implementers or

(01:01:07):
partners are going to win bigin this year.
That's how I am seeing, giventhe way they are able to.
The co-pilot is integrated intoall different areas of Microsoft
M365, you know, licensing,security, data, dynamics 365,
erp, crm, bc and they createdthe platform.
See, one of the things thatSatya did was that we escorted

(01:01:30):
that IDE and GitHub.
All these things are so tiedtogether.
I think this year is going tobe a very good year for
Microsoft.
In fact, in one of the meetingsI think Satya was saying that
he was surprised to see Dynamics365 is winning a lot of bids,

(01:01:51):
you know, in their sales.
I think that's where the powerof this co-pilot agents are
going to be for this enterpriseand implementers, especially
business central space.
If you look at right in the SMBspace, according to Gartner and
Magic Quadrant, there are veryfew competing companies right,
netsuite, bc.

(01:02:12):
So in that also, oracleNetSuite is not having the kind
of ecosystem that Microsoft ishaving right, so they don't have
the Copilot, they don't havethe Azure Cloud.
They don't have the Azure cloud.
They have OCI, oracle CloudInterface.
They are still evolving,whereas Microsoft they have
copilot, their UI and thebackend it's the agents.
I think it is one of the reallygood years.

Speaker 2 (01:02:32):
They can use the connector.
Well, there's a lot to copilotand you're mentioning the
ecosystem from the Microsoftplatform, which to me you
mentioned that fabric is sort ofa blend of a lot of different
separate services now into onelevel plane.
I think the whole ecosystem ischanging as well, where you have

(01:02:53):
the ERP interface, the ERPsoftware, the data backend, the
automation, the tasks.
There's a lot to this.
Copilot within the Microsoftecosystem we talked about
co-pilot studio, we're talkingabout agents, we're talking
prompting when is the best placefor someone to go to learn?

(01:03:16):
And also, how much do you learn?
Right, I mean it's I drive avehicle.
Do I need to know how to buildthe vehicle?
Do I need to know how to fixthe vehicle?
I just didn't know to go totalk to somebody, but I still
need to learn how to drive thevehicle.
Or you know where do you go todo what, but Brad, that's like

(01:03:37):
this journey.

Speaker 1 (01:03:39):
That's like back in the day when someone says I'm in
IT, right.
That's like back in the daywhen someone says I'm in IT,
right.
Back then it was like, oh, youfix computers, but now that's
not the case, right?
When you say I'm in IT, it'slike which part If anyone tells
me they're in IT?

Speaker 2 (01:03:52):
Chris, I don't even talk to them anymore, because if
someone says well, what do youdo?
And they say I'm in IT To meright there, that just means I'm
not going to talk to you,because you think you do
everything and you really don't,because it's impossible to know
everything in IT.
It's almost like AI, because AIitself encompasses much, much

(01:04:12):
more than what people think itdoes with just saying a large
language model.
But where are some?

(01:04:33):
Where can someone go to startto learn their journey of maybe
creating an agent to do emailsfor a podcast, or maybe even
help schedule their calendartime, or I don't even know?
I'm trying to think of all theother practical uses you could
use, both professionally andpersonally, to have some of
these agents simplify your dayso that you have more time to do
the things that are enrichingyou in your life and your
family's lives.

Speaker 3 (01:04:47):
Yeah, sure, I think Microsoft documentation is the
first place to go.
I always refer thatlearnmicrosoftcom there for this
co-pilot.
There is a very gooddocumentation.
Microsoft really improved overthe few years.
That is the best place and toknow complete details, technical

(01:05:09):
details about it.
And if you have a question inthe Microsoft copilot studio and
all those things, go to theforums.
Microsoft is having good forums, dynamics, our Power Platform,
community forum from Microsoft.
I answer a lot of questions inthat, so that's another good

(01:05:30):
place to get quick answers.

Speaker 2 (01:05:31):
Oh, I'm going direct to you.

Speaker 1 (01:05:32):
after this, we're going to have a lot of
conversations.
You do a newsletter too, right?
That's what I was just going tomention.

Speaker 3 (01:05:40):
I publish weekly a newsletter called D365 Co-Pilot
Digest on LinkedIn.
In that, my goal is to justgive a quick update on four
things.
Number one, dynamics 365, erpor CRM.
That is first one.
What are the updates thatMicrosoft is publishing Are they
releasing any new co-pilots?

(01:06:01):
And second one is PowerPlatform how the Power Platform
is evolving, given Microsoft ispublishing are they releasing
any new copilots?
And second one is PowerPlatform how the Power Platform
is evolving, given Microsoft isintegrating AI into copilot,
into their Power Platform source, like Power Apps and Power
Automate.
So that is the second topic Iwrite weekly on the Digest.
And third one is Copilot Studioor AI, microsoft AI related

(01:06:21):
things.
And fourth one is MicrosoftFabric, or, uh, yeah, I,
microsoft, yeah, I relatedthings.
And fourth one is microsoftfabric, microsoft data platform
related.
I feel like these are the corefor this next evolution.
So I write this uh, linkedindigest.
You can find that digest on thelinkedin I.
You know I can share thatinformation we'll also put the.

Speaker 2 (01:06:39):
we have a guest page now, so on the episode on the
website, we'll also put theprofile, which has a link to
your LinkedIn profile, so thatsomeone could read past issues
of your digest and also see thenew issues that are coming out.
So, cy, my mind is blown againthis AI thing.
I don't know where to begin andwhere to end with it, but

(01:07:04):
You're still confused.
It's not that I'm confused, Iunderstand it, you know.
And some things I say in jest,because it's just so much, so
fast, that I think it'simportant to find the nugget
that interests you or you thinkthat you'll benefit from and be

(01:07:25):
aware of the others and pursuethat.
It's almost like the master ofwhat is it?
Jack of all trades, master ofnone.
I think you need to startfocusing on where you think
you'll get the best ROI for whatyou're going to do and then be
aware of the other stuff,because you may have to change
your thought process because ofsomething else that you can

(01:07:47):
incorporate to what you want todo.
So I'm not confused, I'moverwhelmed and exciting, and I
could go down so many differenttangents, because AI itself has
so many different roads that youcan go down side roads or side
streets.

Speaker 3 (01:07:59):
I guess you could say yeah, broadly two, two, two
areas.
Right one is b2c space andunder is enterprise b2b space.
So in the b2c, b2c space thereis so much happening, you're
right who are going to beplayers?
Especially deep seek model isable to.
You know the the thing thathappened.

Speaker 2 (01:08:20):
It's also very interesting the new I need to
get one of those, and we'll havean episode coming up shortly to
talk about that.
But I need to get an lminstalled locally, or do I even
need to right that's my question.
It's do I need to have it orcan I just use one of the
existing tools and models and gofrom there?

Speaker 3 (01:08:38):
I think that's a whole other question it is like
you know your personal laptopversus having a vm on the cloud
right, so do you.
Which one you prefer?

Speaker 2 (01:08:49):
well, it's a matter of mac os well, si, thank you, I
could talk with you for daysand, uh, oh, you're going to
hear from me shortly after thisbecause, chris, we're going to
get some emailing set up,hopefully even some.
Yeah, we should build one out.
Yeah, even even some.

Speaker 1 (01:09:09):
Oh no, sorry, you volunteered did you hear that we
have it on recording?
Definitely.

Speaker 2 (01:09:14):
So yeah, we'll get some emailing ai set up that's
in in some fashion to assistwith the scheduling, to give us
the opportunity to speak withmore guests and not have to
schedule.

Speaker 3 (01:09:25):
Yeah, if you guys are doing manually, we have to
optimize that.
You know, the four-hour workweek book is the best one.
You know, when you try to dothis kind of really at scale,
try to automate it and optimizeit so that you can just move on.

Speaker 2 (01:09:42):
That's where we're looking to go, but again, thank
you again for all theinformation you shared.
Thank you for all that you dofor the microsoft ecosystem.
You share a lot of information.
I read your digest as well, asI do come across you at some
points, uh, within the forum,seeing some of this as I'm
reading up on it.
So we appreciate that, all thatyou do.
In the meantime, if anybodywould like to get in contact

(01:10:03):
with you to learn more about AIand some of the approaches that
are available, what's the bestway to contact you?

Speaker 3 (01:10:11):
Yeah, linkedin, sai Thurlapati.
Or, like you mentioned, therewill be a page.
I will share that informationwith you.
Or D365, copilot Digest is theother newsletter.
Those are the two ways.

Speaker 2 (01:10:25):
Excellent, excellent.
Thank you again.
We appreciate your time.
Look forward to speaking withyou very soon.

Speaker 3 (01:10:30):
Very, very soon, thank you.
Thank you so much.
I listened to your episodes.
I learned so much.

Speaker 2 (01:10:37):
Thank you.
Thank you, we appreciate it.
Have a good day, ciao, ciao.

Speaker 3 (01:10:40):
Bye for now.
You too Bye.

Speaker 2 (01:10:43):
Thank you, chris, for your time for another episode
of In the Dynamics Corner Chair,and thank you to our guests for
participating.

Speaker 1 (01:10:50):
Thank you, brad, for your time.
It is a wonderful episode ofDynamics Corner Chair.
I would also like to thank ourguests for joining us.
Thank you for all of ourlisteners tuning in as well.
You can find Brad atdeveloperlifecom, that is
D-V-L-P-R-L-I-F-E dot com, andyou can interact with them via

(01:11:13):
Twitter D-V-L-P-R-L-I-F-E.
You can also find me atmatalinoio, m-a-t-a-l-i-n-o dot
I-O, and my Twitter handle ismatalino16.
And you can see those linksdown below in the show notes.
Again, thank you everyone.

(01:11:35):
Thank you and take care.