00:03
I’ll welcome to another fireside.
00:05
These are meant to be technology agnostic.
00:08
Mostly I sit here and talk about the industry as a whole.
00:13
You know, my specific realm is within Microsoft 365, little bit of Google.
00:18
00:19
I have some experience with collaboration apps.
00:23
So although this is not specific to technology, I will reference specific technology.
00:29
And today we’re talking about agentic bots versus regular old chat bots in the Microsoft 365 world, they call them the Copilot bots and then the.
00:43
Copilot agents that use that word agents.
00:45
But generally speaking, you might hear that term called agentic AI.
00:50
And what I’m going to explain here is is.
00:54
Mostly universal across like the Salesforce and Google realm and the other people that are doing these agentic and by agentic.
01:01
You can also kind of think of autonomous bots.
01:04
What the main difference is, and I’m going to use up, you know, kind of a scenario a chat bot is.
01:12
Kind of multi-purpose thing.
01:14
So you, you know, you go to your, you go to your scheduling program and you say, what, what do I have on my agenda today?
01:22
That’s, that’s chat bot.
01:23
The chat bot’s going to search your calendar and tell you what calendar events that you had going on for that particular day.
01:30
And it’s kind of just sitting there.
01:33
Until you ask it to do something that’s kind of the key differentiator between a chat bot and then agentic AI.
01:41
Agentic AI is going to act as a.
01:46
Their employee, it’s going to go and take action in the background.
01:50
And based on your parameters, it it may create something that needs your approval or it may create something and you may have given it the full authority or privileges to to take action.
02:03
And one good example of that is.
02:06
I could, I could have a an agentic bot research review my tasks.
02:11
I have a task board, for example, and I have my team and I manage these tasks and I want the agentic bot to tell me if there’s any tasks that haven’t been modified in the last 24 hours.
02:27
Are there, you know, every day, just go ahead and and just look at those particular tasks that that will not just every day, just all the time constantly look at the tasks that I have that are that are late or they haven’t been modified because the presumption is if they haven’t been modified, then they’re probably.
02:47
And in fear of being late and you can have it go even further with predictive analytics, but that’s where we really get to the realm of stuff that I don’t have experience with, but those basic black and white scenarios of if this, then that.
03:03
An agentic vibe can play a really good role.
03:07
And I I heard one podcast from Microsoft where they were talking about technical requirements.
03:12
And this is a really cool, really cool way because I’m a big fan of requirements gathering.
03:18
I’m a big fan of measuring twice and cutting once and wireframing and prototyping and that type of.
03:24
Of due diligence doesn’t always play nicely in the agile realm where we just want to build it and break things fast and move on.
03:34
Of course there’s a balance somewhere between building and requirements gathering, but what I like about the agentic bot idea is that I can sit and consult with.
03:45
You know an individual that needs to build an IT solution.
03:49
I acting as a as a as a consultant, as a software architect, I can sit down and talk with a customer of mine about a business problem that they’re having and provided technical solution by building an app of some sort.
04:04
Well.
04:05
You can sit and talk to them and I, you know, I can mark down their user stories.
04:10
I can, I can use some type of structured document questionnaire to mark down their user stories.
04:15
The personas, you know, you have an admin, you have data clerk, you have a CEO.
04:20
Those are all different personas.
04:22
I take the effort to document out all those.
04:25
Persona is I take the effort to to do some wireframe intake forms, you know some generic wire take intake forms.
04:32
Maybe I do some some relationship diagram, entity relationship diagram models, which is just how the different tables and how the accounts table is going to talk to the customers table if I write that up in a requirements gathering.
04:45
Document and even and even let’s just say I put some rudimentary dashboards.
04:50
So that’s the presentation of the AI document that here is at a 50,000 foot glance not too technical but just enough to kind of show you how this app can look from a blueprint perspective.
05:02
I funnel that Word document into my.
05:05
My SharePoint site and the agentic bot goes ahead and starts taking action it it starts creating based off of the the questionnaire template.
05:14
The agentic app can go ahead and say all right well I’m going to use a power app for this form and it starts building the power app forms based off of this.
05:25
Screens that I’ve built in the Word document, it selects the database, let’s say I referenced a SharePoint database or a data versus database in the document.
05:34
So it goes ahead and provisions that ahead of time.
05:38
I’ve got those different personas and you know that could be based on the the screens could match the personas, so it could start.
05:46
Building those different screens out, maybe it doesn’t connect the a completely but it at least gives me a leg U in getting those screens built, getting that functionality provisioned maybe I mentioned that like a nice presentation power BI dashboard, it starts that template based off some dummy data that it auto generates and the.
06:06
Database, I’ve created it, I’ve told it to create that would be a huge a huge advancement in software development I did see the podcast and they were talking about it creating that scheme ahead of time and if you can have the agentic AI bot follow that particular.
06:27
Man, that is, that is super powerful.
06:29
There’s probably all kinds of different scenarios for your your business.
06:34
You could think of these autonomous spots and another different differentiator between the autonomous spot and the chat bot is usually the autonomous spot is focused on one narrow path.
06:45
You know it’s.
06:46
Just going to do this, I use the the task board.
06:50
It just sees the data on the task board, whereas a chat bot on an intranet site, it’s going to see all the documents and it’s going to be very broad.
07:00
How do I do a password reset?
07:04
You know what, What is the policy?
07:06
What’s the dress code policy?
07:08
It’s just looking for all those documents.
07:10
It may not have in-depth knowledge of a particular area, but the chat bot will be able to find and reference that document based on the large language model O.
07:20
These are some different scenarios in the realm of AI.
07:24
Pilot and all the other fun stuff that’s coming out in in the realm.
07:28
07:29
I you know, like I said, these these principles I mentioned earlier do kind of do cascade relative over to the Google platform as well.
07:38
And I’m sure to a broader extent like the service now and Salesforce world, I have less.
07:44
Experience in those platforms.
07:45
But yeah, let me know how you’re using agentic box.
07:48
Did you know, did you know agentic bots even exist?
07:51
Do you even know how to spell agentic?
07:54
That’s a that’s a good one.
07:55
Talk to you later.
Level Up M365 