The Rise of Agentic AI

Exploring the emergence of AI agents that can take autonomous action in the workplace 

In this second episode of The Macro AI Podcast, hosts Gary and Scott dive into the revolutionary concept of "Agentic AI" - autonomous systems that don't just assist humans but can independently take action, make decisions, and transform how businesses operate. 

What is Agentic AI? 

The hosts define Agentic AI as systems that can reason, make decisions, and act independently within predefined guidelines - contrasting with traditional AI that merely reacts to human commands. Like comparing a simple GPS (which tells you where to go) to a self-driving car (which handles the entire journey), Agentic AI doesn't just provide insights but acts on them. 

Beyond Traditional Automation 

The episode clearly distinguishes Agentic AI from conventional automation: 

• Traditional automation (like RPA) follows strict rules but falters with complexity 
• Agentic AI adapts to changing circumstances, learns over time, and handles unpredictability 

Key Technology Components 

The podcast breaks down the core technologies driving this revolution: 

• Natural Language Processing for Human Interaction 
• Machine Learning & Reinforcement Learning for improved decision-making 
• Knowledge Graphs for Context Awareness 
• API integrations with business software 
• Multi-agent collaboration where specialized AI agents work together 

Real-World Applications 

The hosts explore compelling examples across industries: 

Finance & Banking: 

• End-to-end invoice processing that can verify, route, and even communicate with vendors 
• Real-time fraud detection that prevents issues before they occur 

Healthcare: 

• Automated claims processing that verifies eligibility and medical codes instantly 
• Personalized patient support through AI health assistants 

Customer Service: 

• Support systems that categorize, diagnose, and resolve issues without human intervention 
• Proactive problem-solving instead of reactive support 

Retail & E-commerce: 

• Advanced AI shopping assistants that curate personalized selections 
• Predictive inventory management that forecasts demand and automates reordering 

Human Resources: 

• AI-powered recruitment that generates job descriptions, screens candidates, and schedules interviews 
• Automated onboarding and employee engagement monitoring 

Enterprise Platform Leaders 

The episode highlights how major platforms are implementing Agentic AI: 

Salesforce: 

• Einstein GPT and AI Agents transforming CRM into autonomous business assistants 
• System

About your AI Guides

Gary Sloper

https://www.linkedin.com/in/gsloper/

Scott Bryan

https://www.linkedin.com/in/scottjbryan/

Macro AI Website:

https://www.macroaipodcast.com/

Macro AI LinkedIn Page:

https://www.linkedin.com/company/macro-ai-podcast/

Gary's Free AI Readiness Assessment:

https://macronetservices.com/events/the-comprehensive-guide-to-ai-readiness

Scott's Content & Blog

https://www.macronomics.ai/blog

Transcript

00:00
you

00:12
All right, welcome back to the Macro AI podcast where we explore the cutting edge of AI driven business transformation. I'm Gary. And I'm Scott. And today we're diving into a fascinating shift in the workforce, the rise of AI agents.  Those are agents that don't just assist, but actually take action on their own. This is what we're calling the agentic workforce.  Yeah, it's actually a really interesting term and it's really a concept that's evolving rapidly. I see it.

00:41
almost on a daily basis with a lot of organizations. And if you think about this, AI isn't just answering questions anymore, it's processing transactions, it's resolving issues,  it's collaborating with both humans and other AI agents. Yeah, that's right. And in fact, Gartner predicts that by 2028, one third of enterprise applications will include some form of agentic AI. That's up from pretty much nothing today. So obviously that's a massive shift in the business landscape.

01:10
Can you just imagine what that magic quadrant's gonna look like year over year? uh Usually when Gartner comes out, it changes, especially we saw that in cloud, but I can just imagine the poor person  that has to update that, they're probably almost looking at that on a monthly basis. It's crazy. Yeah, hard to even keep up. I know. I mean, if you think about it, we're already seeing early signs of this transformation in customer service or where a lot of SaaS companies call it customer success.

01:39
seeing it in finance, uh IT automation, and really even healthcare. But what do you think, Scott, it takes to actually build and implement these AI agents at scale? Yeah, that's what we'll cover today in pretty good detail. So we'll break down what agentic AI is,  how it works,  how different businesses are using it, and what companies like Salesforce and ServiceNow are doing to lead the transformation. You see them on TV all the time talking about their new agentic solution.

02:09
Yeah. And we're also going to explore some of the challenges organizations face when adopting this technology, especially with the existing tech death that they have today. So how do they overcome it? So buckle up because it's going to be an eye opening discussion. All right. Well, let's just start with a simple definition. So what exactly is agentic AI? That's a good question. So, you know, when I think of agentic AI, it really refers to  AI systems that

02:38
don't just react to human commands, but take initiative. So these systems can reason, make decisions and act independently within predefined guidelines, which I think is very important to understand. Yeah, that's a huge difference from traditional AI and basic machine learning, which is mostly reactive,  based on algorithms, meaning that it waits for user input before doing anything. Right, right, right. So traditional AI like chat bots or virtual assistants, they really follow predefined scripts.

03:07
but agentic AI can assess situations, it can consider multiple variables and really determine the best course of action on its own. Yeah, I think a good way to think about it is comparing a simple GPS nav system to a, you know, self fully autonomous driving car. The GPS tells you where to go, but you still have to drive a self-driving car. On the other hand, makes decisions in real time and then handles the entire journey for you. That's a little bit more like an agentic AI system.

03:37
Well, it's funny you bring that up because I read this use case a while ago, because I know we're not going to talk about it today per se, but everybody's always worried AI is going to take away my jobs. And great example was  20 years ago,  if you were a cab driver in London, for example, in order to pass the license, you had to understand every single road, street, roundabout, just in the fact that  you may have a traffic jam  and

04:06
Fast forward, GPS comes along. So that's a uh form of AI now. And it's really enhanced  the workflow for those cab drivers. So they don't necessarily have to memorize  all of the routes, although maybe that's still a requirement. But when you think about the technology that AI provides and serves  and working alongside humans, it's really going to be a lot of  the content that we talk about in  future episodes. ah And if you think about in the business world,

04:36
You know, that means that AI doesn't just give you data insights, but  actually acts on them just like GPS.  Uh, you know, so ways, example.  Um, so similar to approving travel requests or processing invoices or, uh, even negotiating contracts with suppliers, know, those are, those are things to think about from that perspective. Yeah. Processing invoices.  So, yeah. So let's just talk about automation. So automation has obviously been in business for decades. So.

05:04
Let's just kind of chew on what makes agentic AI different, which is, what's, what's happening right now. Yeah, that's a question. So traditional automation, like robotic process, optimization, everybody knows that is RPA follows strict rules. So if X happens, you do Y it's efficient for repetitive structured tasks, but falls apart when things get complex or unpredictable. It just doesn't, it doesn't function the way that you would expect kind of doing something ad hoc.

05:34
Yeah. People think of robotic process automation, they think of robots, but it really, there's a lot of solutions that are considered to be RPA. So RPA is like an assembly line worker. It's great at repetitive actions, but it's completely useless. if something unexpected happens, it can't react to anything unexpected. Yeah, exactly. So in comparison, agentic AI  is more like a smart executive assistant. It can analyze different factors.

06:02
adapt to changing circumstances and make decisions based on real time data. Yeah, I think  customer support's a good example. So, you know, the simple chat bot that you interface with all the time, customer service, it might respond to FAQs that are uh in a simple database, but an agentic AI system can actually diagnose issues,  escalate urgent cases, or even schedule a technician visit all without any kind of human intervention based on the inputs that are given to it.

06:32
the signals that are given to it so it can react. Right. So that's pretty powerful because you can, especially if you're in that type of environment, know, everybody talks about following the sun in terms of support, but you can really deliver that 24 by 7 support in that agentic and AI environment. And I think that's the part of the key here is the autonomy and adaptability. So these systems don't just follow a script, you know, they're learning and evolving over time. Right. I think that's kind of what we're saying here.

07:01
Yeah. I mean, and you can really predict that in the future, a lot of executives are foreseeing that they're going to have full teams  of  agents that are out there doing different tasks for them. So this is where things get really interesting. We hear terms like AI agents  and agentic AI used interchangeably, but they are actually different. So I think we need to break that down a bit. Yeah. So actually really good, important distinction, you know, an AI agent.

07:28
is just a single piece of software designed to perform a task like an AI that schedules meetings or suggests the best route for deliveries, you know, kind of like we were talking about before. Exactly. But agentic AI  is the orchestration of multiple agents working together.  It's a system where different agents collaborate to handle complex workflows. So it could be in any business workflow, or you might have a team of agents,  you know, working for your department. Okay. So, so for example, if you ran some sort of

07:58
e-commerce platform, an AI agent might recommend products to customers, but in an agentic AI system that would track inventory, process orders, uh handle customer complaints,  probably even negotiate with suppliers like we were talking about before, all really in one seamless workflow. Yeah, that's obviously a huge leap forward, but that's what's happening now. So instead of isolated automation tools, we're talking about

08:24
an intelligent ecosystem of AI agents that can truly manage operations. And a lot of big companies, software companies, SaaS companies are starting to build products that are using  AI agents in a full intelligent ecosystem. Okay. Yeah. So we should probably dive more into the tech behind agentic AI  and  understand how it really works. maybe Scott, you can kind of walk us through that. You've seen quite a bit of this. Yeah. Jump into a few of the...

08:53
So one of the key ones that a lot of people are familiar with is natural language processing or NLP, also known as natural language understanding gets a little bit deeper. And so that's where AI understands and interacts with humans using natural language. ah Then there's,  you know, machine learning and reinforcement learning that goes behind that. the AI agents improve their decision-making over time. That's a key component. So they learn from what they've the activities that they've been performing, and then they improve

09:22
how they respond over time. ah Then in the data space, you've got knowledge graphs and context awareness. So that's where agents  can remember the types of interactions and adapt using knowledge graphs and context awareness.  And then obviously they need to connect to all these workflows. So kind of the core underlying matrix are the APIs and the integrations so that AI agents can connect to business software like Salesforce and ServiceNow. Yeah, that makes sense.

09:52
So  we probably don't want to forget about the multi-agent collaboration. So instead of one AI trying to do everything, different specialized agents work together. Some handle the data, others execute the tasks,  and an orchestrator agent  oversees everything  through this entire workflow, right? Yeah, the orchestrator agent. Yep. ah So that's what makes agentic AI scalable. It's not just one AI, it's an entire network of smart agents handling different parts of the business process.

10:22
Yeah. So if I think of real world examples of agentic AI, you know, where is agentic AI already making a difference today in your opinion? Well, there are a lot of industries being impacted and transformed by it. So it's, you know, it's really just starting now, but, know, let's, let's go one by one. Let's just break down, say finance, for example.

10:45
ah So let me think about this. So finance and banking. So autonomous invoice processing and fraud detection. How about we tackle that one? So yeah, I mean, because that's just, it's just a  key component in that industry. So if you think about this traditionally invoice processing is a nightmare for businesses. ah I don't know any business that really says I just love invoicing. All right, maybe you do, but I haven't come across many, but it's  the key to

11:15
revenue from an accounts receivable and also uh making sure your expenses are paid from an accounts payable standpoint. But it really involves a lot of manual entry, back and forth approvals,  often takes days or even weeks to complete. And sometimes there's errors from the supplier and errors from possibly you, you know, going out as well. uh So, yeah.  So think about the impact on the CX or, know, the customer experience. So

11:42
Mistakes in invoice processing are going to cause delay payments, frustrated vendors, issues with compliance. But with agentic AI, the whole process can be automated end to end. And you can have a human in the loop to do inspection where you think it might be necessary. Okay. So if I'm a CFO, the first thing they're going to ask is how does it work? Because if I can ensure that my receivables come in,

12:08
one or two days faster. That's, that's huge for my bottom line. Right. Yeah.  So think, think about, um, you got hundreds of invoices coming in every day instead of having employees manual manually sorted through them. An AI agent can instantly scan the invoice, extract  specific details that you need, like vendor name, amount, due date, whatever. And then it can validate it against company records. Yeah. So if there's a mismatch, like a pricing error or missed purchase order,

12:36
the AI agent doesn't just flag the issue, actually reaches out to the vendor automatically requesting a correction or additional information.  having a manual person having to  then go find it and then start the process. Yeah, 100%. So once  it's verified, the agent can route the invoice for approval, uh obviously following any predefined business rules. And if it meets all the criteria, it can auto-approve.

13:04
process the payment all without human intervention. Yeah. It's a huge, huge, huge efficiency boost. so what about fraud detection? Yeah. So  fraud detection, a  pretty critical topic these days. So agents can analyze transaction patterns and they can do this in real time based on the processing power. So if an unusual transaction occurs, like a huge invoice or payment sent to an unverified account, the system can flag it.

13:33
Pause the transaction until further review, know, human in the loop again. Yeah. Or even probably conversely,  you know, where you've seen credit card fraud in the past, it's these small, you know, $1 50 cent charges just to, to ensure that they can get things through. And to your point, if it's not an approved vendor, it may, may be able to flag that as well. If, you know, taught by the organization to, look for those types of patterns. So instead of finance teams,

14:02
really chasing after the fraud, after it happens, know, agentic AI can help prevent that real time. So that's obviously a real game changer just from a cycle time of having to, you know, request a charge back uh from the banking financial institution. And, you know, so definitely  is, is a huge game changer in my opinion. Yeah. So let's,  let's pivot over to healthcare. So obviously healthcare, huge industry, what going on there.

14:31
uh a lot of opportunity for AI. So, you I think one of the keys there would be faster claims processing and really personalized patient support. So those two things are pretty critical.  So, so yeah.  I'm agreeing with you. Sorry. I didn't mean to interrupt. Yeah, go ahead. Yeah, I was going to say, so when a, when a patient submits a claim, typically goes through, I feel like 20 different levels of light within the

15:00
healthcare organization. So it's verifying eligibility, checking against policy terms, ah what else, approving  or denying it,  which I feel like usually falls to the latter, denying it. ah So this really can take days and sometimes weeks. And I can personally attest to this. I received a bill, ah I thought I had a concussion two years ago and I finally just received the invoice  literally last week. Crazy.

15:29
Yeah, it takes a long time.  with a gentic AI, can get all those processes moving simultaneously. So it's the whole thing is just completely rebuilt. So the agents can extract all the details, reach out and verify against insurance policies, cross check all your medical codes instantly and just,  move the whole process along, you know, end to end. Yeah. So even if the claim is incomplete.

15:52
Yeah. So yeah, so yeah, if it's incomplete, the AI agent can automatically request additional documentation from the healthcare provider or from the patient without waiting for a human to step in. can call out, they can use a number of different media to reach out to get that information. So obviously we've all probably experienced an issue or an error or a question. So what about personalized patient support?

16:15
That must be a game changer there.  Yeah. Personalization is what people are expecting now. So that's, that's a great use case. So AI powered virtual health assistance,  uh, they can help schedule appointments and we're going to start seeing this pretty soon. lot of DSOs, dental service organizations are just starting to do this now. You're going to see it across the board, but it can remind them about obviously the appointments. can remind them about medications. can provide basic medical advice,  basic medical advice based on their history. Yeah. I think that's, that's important because

16:46
Instead of the traditional generic chat bot where you're talking about AI the way it was designed before, it now understands the patient history. It's not just giving you these just general canned responses when you want to customize something a little bit more. So it has more of a conversation interaction with the end user, in this case me, if I'm calling in on that bill. it's basically taking action based on real needs, not regurgitating six different

17:16
topics in a chat bot. Yeah. And that's just one sliver of one use case in healthcare. Lots of them out there. Right. Exactly. So then if we shift gears to customer service or customer success,  this is an area where agentic AI is already having a massive impact. You're probably seeing this  on your side of the business today. Yeah, no doubt. And you and I have been doing this for a while with customer service.  you know, traditionally if a customer has an issue, you know, say their internet is down.

17:45
you that's happened a few times, especially to me recently. yeah, so  can call into a support line, you're on hold, then you explain the issue to somebody that may or may not understand what you're talking about. Then you get transferred to another agent that may or may not understand what you're talking about. And then maybe you get your problem solved. Maybe. But with agentic AI, that's not the case, right?  Right. So with agentic AI, as soon as somebody reports an issue, an AI agent can

18:15
categorize the problem,  check the database for solutions, check routages in the known area,  a whole list of things that it can automatically check and they can either resolve it instantly or escalate it to the right department,  hopefully the right department. Yeah, exactly. So in your case, if it was an internet issue or some sort of service issue or something that you were under contract for, the AI agent can essentially schedule uh a tech visit, send real-time updates to you as the customer.

18:44
and even follow up afterwards to ensure the problem was fixed. you know, whether it's  closing the loop there, but also doing a survey and  improving that MPS score for customer success or customer service, right? Yeah. The difference now though, is that with artificial intelligence and the power of these processors, the database of potential problems and technical detail can be much more expansive than the chat bot that you're used to seeing when it pops up on your computer screen.

19:12
So  next generation customer service is AI powered. Yeah, absolutely makes sense. Plus just the amount of customers you can cover with the tool versus  adding additional  staffing that understands every single complex issue that could potentially happen just from a training and LMS standpoint is  really powerful. So then if we talk about that, so if we build into that,

19:42
And we think about retail and that e-commerce topic we were mentioning earlier. We've all used  recommended engines before, but how, in your opinion, is agentic AI taking things to that next level  in this type of environment within retail and e-commerce? Yep. Good question. So I think the shopping experience is going to completely change. So imagine shopping online and instead of browsing through hundreds of products, an AI agent asks what you're looking for and curates a personal

20:11
selection based on your preference,  your past purchases, uh your chat history, and even current promotions. Great. instead of who bought the, people that bought this also bought that, which I see all the time when I'm, if I'm shopping online, it's really a true personal shopping assistant in this case. Yeah. Yeah. And say, for example, something's out of stock, your AI agent  can go check other

20:38
online e-commerce vendors or stores or whatever. And it can even predict restocking timelines, pre-order for you, lot of different things that this AI assistant can do for you. That's pretty impressive. go a little further into inventory management because this is a pain point for one of the retailers that I like to purchase things from. We won't say who they are, Yep. Yep. Yeah. So on the operational side, AI agents can, they can constantly monitor stock levels, you

21:08
across the organization. can track sales patterns and they can automatically reorder products before they run out.  Then they can predict demand spikes like increased sales of winter clothing uh before a cold front hits. But not only that, you could even make them even more intelligent to predict uh when the cold front would hit based on monitoring weather services. Yeah, because all that data is available real time.  if they're able to ingest that.

21:35
and make some predictions, maybe the low state of deviation of that event,  snow, rain, ice, uh it makes sense. So it's really creating a smarter business  for the organization, but also their end customers. So instead of just reacting to an inventory issue, it's out of stock, companies can stay ahead of that demand in real time, which absolutely makes sense, especially in a thinner margin type of vertical

22:04
such as retail. Yep. Definitely pretty amazing. ah Let's just wrap it up with human resources. mean, hiring, for example, one of the most time consuming tasks. How can agentic AI help in that process? Yeah. So I think of it this way. ah If you look at the traditional hiring process,  HR, chief people team, hosts a position, they screen resumes.

22:33
They schedule the interviews and conduct assessments and finally make an offer. I know I'm probably skimping over that a little bit for some organizations, but we all know that that can take weeks if not months. Yeah. Yeah. So,  so with agentic AI, if you have an agentic AI system in HR, it can do everything much faster. So obviously it can generate job descriptions based on the company needs. uh

22:59
It can screen resumes instantly, ranking candidates based on skills and experience, and then it can reach out to them and schedule interviews automatically by checking availability for the recruiters and reaching out to the candidates  via different media. ah And then it can even conduct AI-powered initial assessments. So it could pull up a video and have a video interview. uh And it can even  monitor for sentiment analysis for your potential candidate.

23:27
So  you'd essentially be in a conversation,  say on a video platform potentially, or even just an audio platform with an AI agentic uh assessor  of that particular machine. So that's pretty cool.  then- Yeah, that's happening.  Yeah. So then after that, let's talk about what you're seeing after the person the candidate is hired.

23:54
Yeah. So that's, that's where it gets a little bit easier. So AI agents can help onboard the employees so they can assign certain training modules based on their skillsets and job that they're hired for.  Uh, they can, they can monitor for employee engagement. making sure that, uh, you know, everybody gets the support that they need along the way through the process. Yeah. So it's really a complete  HR transformation for many organizations and probably assists companies that are smaller.

24:23
whether you're even in an early stage company and maybe you have somebody that's in charge of HR, but asking them to  help staff and grow the organization. know a lot of sales  organizations look for that pretty quickly. So the CRO, early stage company, or maybe has initial funding, they're looking for particular candidates  and there may only be one person on the HR side. So instead of HR teams,

24:50
you know, or that HR person getting buried in paperwork, they can focus on strategy and the people that are coming in and out of the organization, but also utilize the technology to allow them to cover more ground and allow them to scale a lot faster. Yeah, exactly. I mean, that kind of sidebars us into another topic that  I think because of AI, they're going to be all whole realm of startups across every industry.

25:15
And this obviously would allow you to have a really nimble, thin HR team, as long as your agentic system is working the way that you need it to be designed. Yeah. So, yeah. So just to wrap it up, that was a lot of ground that we covered. So just to recap, so agentic AI is revolutionizing finance, healthcare, customer service. We also talked about retail and then finally HR. And this is just the beginning.

25:42
Yeah, I  this is just the, would say phase one, especially with the ability to interact with the human on the other end. Obviously  there's going to be continued learning from the machine itself  on the uh agentic side, but that's part of, and we'll talk about this in future episodes, you what does AI mean and kind of the different categories. There's a lot of supervised learning that can continues here, but this is

26:12
an excellent use case of how AI works along with  humans in the day-to-day job.  Totally agree. I think what's happening now is that  most people really don't know that much about the power of AI. Until you're really into it, it's really hard to know the full capabilities. But the business leaders that start to embrace this now, they start to embrace learning and they start learning about use cases, they're going to be way ahead of the curve as AI continues to evolve. Great recap, Scott.

26:41
We'll be back here shortly on future episodes.  If you want to find us,  both Scott and I will have our LinkedIn profiles  in the show notes.  And if we have any additional links to put in there that you can always find that in the notes section as well. But reach out to us.  We'd love for you to subscribe and let us know other topics that are interesting for you either to learn more about or areas that you're looking to explore as Scott and I deal with

27:10
many companies on a regular basis just on the AI journey. So we'd be happy to chat. So until next time, thanks for joining. Appreciate the time today, Scott. was great catching up with you as always. Yeah, good one. Talk to you later. All right, see ya.