#348 The Future of Automation: Agentic vs Deterministic Workflows

Summary

---

Podcast Summary: Embracing Digital Transformation

Episode #348 — The Future of Automation: Agentic vs Deterministic Workflows

Host: Dr. Darren Pulsipher
Guest: Sid Brott, Founder of Refound
Date: May 5, 2026

---

Episode Overview

This episode dives deep into the evolving landscape of automation, contrasting traditional deterministic workflows (like RPA) with the rising wave of agentic, AI-driven processes. Host Dr. Darren Pulsipher welcomes Sid Brott, an entrepreneur and longtime AI practitioner, to demystify these terms, explore real-world implementation challenges, and look ahead at how enterprises can leverage hybrid or heterogeneous workforces blending people, deterministic automation, and reasoning AI agents. The conversation is practical, jargon-light, and focused on actionable insights for public sector and enterprise listeners grappling with digital transformation.

---

Sid Brott’s Origin Story and Evolving with AI [(01:22)-(05:07)]

• Sid’s background: Began as an engineer, moved into consulting, then entrepreneurship in tech, marketing, and product management.
• Early AI experience: Worked with OpenAI before mainstream adoption, built one of the first companies on the GPT-3 API in 2020.
• Timing and growth:
  • Sid notes, “It was so early that the models weren't all that great… companies ended up saying, ‘we could probably create better content than this’” [(03:44)].
  • The platform’s capabilities rapidly improved, with Sid observing the difference between the slow start and sudden leaps: “It's very hard to know. Like when you're in an exponential curve… at any point it always looks like a linear curve” [(05:48)].
• Transition: Leveraged early-mover status to build ventures as models matured, and now leads Refound, a consultancy focused on agentic automation.

---

Automation: Deterministic vs. Agentic Workflows [(08:55)-(13:20)]

Deterministic (RPA, Traditional Automation)

• Definition: “Your typical RPA and automation in general is [a] very predetermined workflow… if this happens, then do that… so you can automate a lot of processes like this.” — Sid Brott [(09:16)]
• Use cases: Best for processes with clear, invariant rules and few exceptions.

Agentic Workflows

• Definition:
  • “The agentic workflow is where there's a lot of variance in what's happening that you can't just deterministically say if this happens, do that… when a workflow has more exceptions than rules.” — Sid Brott [(09:16)-(10:06)]
  • Involves reasoning models able to evaluate unique, unscripted situations and determine the best course of action.
• Example: In customer support, standard requests (“where is my shoe?”) are deterministic; handling complex, ambiguous refund claims requires agentic reasoning.

Memorable Quote:
“I love how you said that—a workflow that has more exceptions than rules. Absolutely, I get it. That makes sense.” — Dr. Darren [(10:06)]

---

Blending Approaches: Hybrid/Heterogeneous Workforces [(13:20)-(16:33)]

• Systems In Practice:
  • “You can just have questions that are very basic go straight through regular automations, and then questions that are more complex go to a reasoning model.” — Sid Brott [(13:00)]
• Trade-offs:
  • Agentic workflows can be less predictable and possibly costlier per transaction, but dramatically reduce human workloads by automating previously “un-automatable” exceptions.
  • “Maybe 20% of the questions are being automated and 80% handled by a human. For agentic systems it’s 80% handled by the AI and 20% by human—or maybe more.” — Sid Brott [(13:46)]
• New workplace paradigm:
  • Dr. Darren suggests “heterogeneous workforce” as more accurate than “hybrid”: “I've got humans, I've got agentic flows, and I've got deterministic workflows... This is actually pretty cool.” [(14:29)]

Case Study:
Nike customer refund — traditional system sends all exceptions to a human. Agentic AI now evaluates customer history, order value, and can recommend, for instance, a 70% refund with a final human review only for confirmation.

---

Orchestrating The New Workforce [(17:39)-(18:29)]

• Workflow architecture:
  • A system now must route each case to deterministic automation, agentic AI, or a human.
  • Could be “an AI orchestrator, or...basic AI classification that just looks at what's coming in and classifies [the request]” — Sid Brott [(18:12)]

---

Data Privacy & Deployment Concerns [(18:29)-(21:07)]

• Enterprise & Public Data Security:
  • Options include using enterprise versions of OpenAI, Anthropic, Gemini, etc., which offer compliance (HIPAA, regionally located data centers, no data used for training).
  • Open source models (Meta, Google Gemma, etc.) can be deployed on-premises if maximum privacy is necessary.
  • “Open source technology is nearly as good as the paid.” — Sid Brott [(21:00)]
• Listener Scenario: Real-world patient info example, confirming privacy-compliant AI usage is possible.

---

What Makes Agentic Different [(22:00)-(23:56)]

• Classic chatbots vs. agentic AI:
  • Traditional ChatGPT workflow: The user provides context and executes actions; AI just advises.
  • Agentic workflow: The AI gathers context, does research, makes recommendations, and can execute actions; humans review or approve.
  • “You tell the AI what to do. The AI does the work. That's the easiest way of working.” — Sid Brott [(23:14)]
• Power shift: “We're in charge with the agentic flows. We're in charge, and we need to make sure that we stay in charge.” — Dr. Darren [(22:58)]

---

Implementation Challenges & Organizational Change [(24:33)-(26:23)]

• Integration hurdles:
  • Major challenge is unifying data across silos and systems (“the biggest problem is ...so many different data silos...to first get all of that data into a common data context layer...”) [(24:46)]
  • Requires deep process understanding, mapping tools/integrations before AI agents can be built.
• Data access/security:
  • Solutions run on the client’s infrastructure: “We build the agents into their existing servers...so it sits within their [data], within their servers and their architecture.” [(25:54)-(26:23)]

---

Consultancy & Change Management [(27:13)-(32:38)]

• Practical engagement:
  • Start by diagnosing bottlenecks and workflow pain points.
  • Process is akin to classic management consulting (“good old fashioned process consultant,” jokes Dr. Darren [(28:53)])
  • Sid distinguishes their model: Instead of costly headcount-based solutions, they deliver automated workflows (agentic flows) that reduce recurring labor costs.
• Training & hand-over:
  • Clients are trained to use and maintain the systems, with technical support available for custom changes.
  • “If we've understood the process well enough...there's really not much else to do.” [(30:56)]
• Business model future:
  • AI agents as scalable, repeatable resources (potential “AI resource rental” idea), but with necessary customization per client due to unique business processes.

---

Getting Started and Practical First Steps [(33:10)-(34:49)]

• Sid’s exercise for listeners:
  • Map out the entire workflow of a key process.
  • Identify what must remain human-led and what parts could be handled by agentic AI.
  • Focus automation on the most repetitive, time-consuming, or exception-heavy steps.
  • “Everything else that could become AI agent...especially looking at the parts that take the longest or [are] most repetitive, those are the lowest-hanging fruit to automate.” — Sid Brott [(33:54)]

---

Notable Quotes & Memorable Moments

• Sid Brott (- on exponential growth):
  “When you're in an exponential curve at any point it always looks like a linear curve” [(05:48)]
• Sid Brott (on agentic workflows):
  “Workflow that has more exceptions than rules” [(10:06)]
• Dr. Darren (on hybrid workforce terminology):
  “Heterogeneous workforce ... because I'm using more than two a lot of times” [(14:29)]
• Sid Brott (on paradigm shift):
  “You tell the AI what to do. The AI does the work.” [(23:14)]
• Sid Brott (on real deployment):
  “The biggest problem is … so many different data silos … to first get all of that data into a common data context layer” [(24:46)]
• Sid Brott (best first exercise):
  “Map out the whole process ... every step of the way, map it out ... decide after you map out the full workflow what should be human, what can go into AI ...” [(33:19)-(33:54)]

---

Key Timestamps

• [01:22] — Sid’s AI & entrepreneurial origin story
• [08:55] — Deterministic automation vs. agentic flows
• [10:06] — “Workflow with more exceptions than rules”
• [13:20] — Blending deterministic and agentic approaches
• [15:06] — “Heterogeneous workforce” concept
• [18:12] — AI orchestrators for workflow routing
• [19:09] — Data privacy & open source LLMs
• [22:00] — What makes agentic AI workflows truly different
• [24:33] — Deploying agentic flows: real barriers
• [28:53] — Consulting parallels
• [33:10] — First steps for automating your own processes

---

How to Get Started / Learn More

• RefoundAI: Visit refoundai.com to engage with Sid and his team directly.
• Action item from Sid: Map your process, identify automation candidates, and start with high-friction repetitive steps.

---

Tone & Takeaways

The conversation is candid, approachable, packed with practical scenarios and analogies to ground complex new technology in reality. Listeners will walk away understanding:

• The clear distinction between rules-based and reasoning-based automation.
• Why agentic workflows are just now becoming viable—and why most organizations aren’t operationalizing them (yet).
• How blending deterministic, agentic, and human-driven work is forming the backbone of the future enterprise.
• The practical path to start their own transformation.

“You tell the AI what to do. The AI does the work.” (23:14) — That’s the future of digital transformation.

---

Transcript

[Emergent Narrator]

Before we get into today's episode, a quick thought. Right now, everyone is chasing the next big thing. AI, automation, agentic systems. And yes, every wave creates opportunity. But here's the problem. What feels like an advantage today becomes the baseline tomorrow. So the real question isn't how do you keep up, it's how do you stay ahead? That's why Emergent exists. For over 17 years, they've helped organizations build a better way of working, what they call a work operating system, connecting people, process and technology. Because the only sustainable advantage is how your business actually works. Get ahead, stay ahead. Learn [e-mail]

[Dr. Darren]

welcome to Embracing Digital Transformation, where we explore how people process, policy and technology drive effective change. This is Dr. Darren, Chief Enterprise architect, educator, author, and most importantly, your host

[Emergent Narrator]

on this episode, the Future of Automation Agentic versus deterministic Workflows with Sid Brott, founder of Refound.

[Dr. Darren]

Sid, welcome to the show.

[Sid Brott]

Thank you so much for having me, Darren.

[Dr. Darren]

Hey, I'm really excited about this topic because I've been playing around with AI agents and agentic workflows and things like that, but you're going to school me today on the proper use of the terms and what it means and how to really use these to operationalize a business. But before we get started, everyone that listens to my show knows that I only have superheroes on the show. Every superhero has a background story and some superpowers. We'll uncover the superpowers. But can you give me your background story, your origin story?

[Sid Brott]

Absolutely, yeah. So I've been in the AI space for many years, actually like eight or nine years. But my career started long before that when I started off in tech as an engineer, got into consulting and then got into entrepreneurship from there to start companies, found my way into, like, marketing and product management, wound up working with some Silicon Valley companies. And then in 2019, I heard about this tiny little startup that no one had heard of. It's called OpenAI. And they were working on these cool things called language models. And so I was also working with some AI startups down in the valley at that point as sort of a marketer for hire kind of thing. And then in 2020, I came back to Canada when the pandemic hit. And at that point I said, well, I want to start my own company. I've been working with startups my entire career. I want to start something. And because I'd got that interest in AI in Silicon Valley, I started to play with these models. And if you have been in the space for long enough you may have heard that OpenAI launched GPT3, which was the first kind of publicly available.

[Dr. Darren]

Yeah, November. November 30th, 2022. I know the date.

[Sid Brott]

No, that's not the one. No, that was ChatGPT.

[Dr. Darren]

Oh, you're right, you're right. That was ChatGPT. You're right.

[Sid Brott]

That's when most people had heard of AI. I was working on it in 2020 when they launched the API. Just the API.

[Dr. Darren]

Just the API. Gotcha.

[Sid Brott]

To a small group of people. I was lucky to get in on the early list there. And we were the first to launch a company built on that API. It was an AI marketing software.

[Dr. Darren]

Very cool, Sid. You hit the market, right? That's very cool.

[Sid Brott]

I think it was too early. So we did raise funding, but it was so early that the models weren't all that great. And the companies that used us ended up saying, well, we could probably create better content than this. And so we had to shut it down. And then of course, in November 2022, that's when ChatGPT came out, which used a newer version of the model and AI exploded. And while I was already well positioned to take advantage because I'd been in the space already for like three years at that point. And so, yeah, I got into a venture. The VC fund that had invested in us brought me on to run the Aventure Studio. So we built a bunch of AI companies and now I run a consultancy called Refund.

[Dr. Darren]

That is. That is really cool. It's really cool to talk to someone that was there at the birth of all this, which is really cool because you. You hear about the gray hairs like myself that were there at the birth of the dot com boom in Silicon valley in the 90s. You're going to be famous, you know, to your kids and grandkids. Someday you're going to sit around and say, oh, I remember when you know all those great things. So this is awesome, Sid.

[Sid Brott]

Yeah, yeah, it was. I mean, I think I was just like very passionate about it and have always been passionate about tech and sci fi kind of stuff and AI ever since I was a kid. So I'm glad to be in the space right now.

[Dr. Darren]

That's awesome. All right, so with this incredible background you got in at the ground floor, things were very. What's the right word? Rough at the beginning, right?

[Sid Brott]

Yeah.

[Dr. Darren]

4K token context window. That's not a lot to play around with. Yeah, it hallucinated like crazy.

[Sid Brott]

Oh, my God, Yes.

[Dr. Darren]

Yeah, Right? You've seen huge changes and were you expecting the changes that you saw so quickly or were, were your expectations that high that, hey, it needs to get to this point in order for it to be good enough?

[Sid Brott]

You know, I, I, I did not expect it to get to where it is today. You know, like back in 2020 when we were first playing our first of all, I thought it was magic in a box when I used the API and I asked it, what is the meaning of life? And it gave me this beautiful answer. And it was very like, well written and coherent. But of course that was like, you know, when we were trying to use it for creating marketing materials for building a startup, it wasn't always that the best, you know, And I thought, okay, well, it would sort of like improve over the years. I just didn't know, I didn't realize. You know, it's very hard to know. Like when you're in an exponential curve every, at any point, it always looks like a linear curve. So in the early days where it was kind of like the flat part of the exponential, it just didn't really, I didn't realize, I didn't even think, I didn't think it would get to this, where it is today, so fast, at least.

[Dr. Darren]

Well, see, that's really interesting. I like how you explain the exponential curve because you're right. It's very, the slope is very flat at the beginning of an exponential curve. So you rode that all the way through. So it, it surpassed your expectations from the, from the initial. So are we, are we at the top of that exponential curve or are we starting to head down now? Is it the, the hype cycle curve or are we still heading up? What do you think?

[Sid Brott]

I think they're still heading up. I just, I don't, it's, it's, there's, there's the, it felt like there was a bit of a slowdown maybe kind of in 20, early 25, late 24, but then there were a couple of step changes that happened. So in kind of like spring of 25 and then four months ago, there were a couple of like big jumps in capability. Especially four months ago, where you had open Opus 4.6 and some of the

[Dr. Darren]

other reasoning models really started taking off then.

[Sid Brott]

Exactly. And the agentic kind of workflows started to become viable. It went from like, it would make too many mistakes to be in production to it doesn't, it randomly makes mistakes. And if it does, then maybe there's something wrong with like the context of giving it. Yeah.

[Dr. Darren]

All right, so we're, we're at, we're at a point now where I can really start using these with the gentic flows is what I'm hearing.

[Sid Brott]

Absolutely. Yeah. I mean that's what our company does. We deploy it into a. Organizations, enterprise organizations in production, and they use those agentic workflows in their work.

[Dr. Darren]

Let's pause for a second.

[Emergent Narrator]

Most consulting creates dependency. More frameworks, more slides, more reliance. Emergent takes a different approach. They believe consulting should leave you stronger than it found you. That means solving real problems, building internal capability, and helping your teams actually change how they work, not just talk about it. From process to leadership to technology, their focus is fix the way work gets done. Because when that changes, everything else follows. Learn [e-mail].

[Dr. Darren]

Okay, so let's talk about the difference between an automated like rpa, automated workflow and an agentic flow. Because I think let's dumb this down as dumb as we possibly can make it so that people can understand. Can you help me out with that a little bit?

[Sid Brott]

Sure. Your typical RPA and automation in general is very predetermined workflow, right. With like, so it's deterministic. You say if this happens, then do that. If this happens, then do that. Right. So whatever conditions or triggers or whatever happens, first do this, then do that, then do xyz. So like you can automate a lot of processes like this. And oftentimes, you know, when people come and say, hey, can I use AI to automate this? And like you could possibly just do a basic automation. Now the agentic workflow is where there's a lot of variance in, in what's happening that you can't just deterministically say if this happens, do that. Because there's so many different conditions, so many variables that it's hard for you to know what could happen when there's, you know, it's, it's, it's when a workflow has more exceptions than rules, basically.

[Dr. Darren]

Okay, I love how you said that a workflow that has more exceptions than rules. Absolutely. I get it. That makes sense.

[Sid Brott]

Yeah, because that's, at that point you have the agentic piece of it come into play, which is an AI model that has reasoning. So a lot of your day, models that start coming out from like early late 24, early 25 are reasoning models where they're able to think through a question or a problem and then come up with like a solution. And so what's happening now is for a non deterministic workflow where there are more exceptions than rules, what it does is it looks at what's coming in and then says, hmm, what should I do here? What would be a reasonable choice to make here? And then it makes that choice and goes ahead and does the rest of the automation. Does that make sense?

[Dr. Darren]

Yeah, it does. I like that a lot. It still has a goal it's trying to achieve. It's not coming up with the goal on its own.

[Sid Brott]

Right.

[Dr. Darren]

It's been told, hey, given this input, your goal is to do this over here. It's not creating its own flow or its own goal at the end.

[Sid Brott]

Right, yeah, yeah, exactly. So we can take a common workflow that's, you know, that has previously had a lot of RPA and regular automation and now is benefiting from agentic automation customer support. Right. So you think about when a customer likes say like Nike's customer support. Right. Oftentimes a lot of the questions are coming in is like, hey, where is my shoe? Or what is the refunds policy? Things like that. Those are very basic. And then you can obviously just automate that by just matching like the query, the question where's my shoe? Or similar variations to an answer. Maybe goes and checks the no order system. The order system and gives you the answer. Right. So that's a very. That you don't really need agentic AI for, but you do need agentic AI for a whole host of all the other questions that lies outside of that. Right. Like it could be, hey, I was running with this shoe yesterday and there's a small little mark and what can I do here in this case? Can I send it back? Or whatever Complex question right now you need an agent to think through. Okay, does this merit a refund? Does it merit just sending them to a help center? Does it merit sending them to a customer support person or whatever. Right. There's like so many different things that they, that actions that could take and you have to determine which one to take. And that's where the agent can come in and like process the question and then figure it out.

[Dr. Darren]

Okay. I love how you distinguish between the two. Do both workflow and agentic flow. Can they both sit in the same system together?

[Sid Brott]

Yeah, yeah they can. Right? So you can just have like questions that are very basic, that just goes straight through regular automations and then questions that are more complex that goes there, that gets sent to a reasoning model that tries to think through the answers and then gives you the appropriate answer or takes the appropriate action.

[Dr. Darren]

Okay, I like that answer. Because sometimes you want things highly deterministic, right? Because agentic flows are non deterministic. It means the outcomes aren't as determined either. Right. It could make a mistake.

[Sid Brott]

Exactly. Yeah. Yeah.

[Dr. Darren]

Okay, so. And then also it could make more. It could take more time and cost more to do an agentic flow than like a workflow, potentially.

[Sid Brott]

Yeah, exactly. Well, like, you know, those, those costs are coming down, but just. Yeah, sort of like in. In large volumes then. Yes. Yeah. But again, like, the other thing is, if you think about the regular automation, it could only handle certain types of answers and everything else would go to the human. So if you think about that system, maybe 20% of the questions are being automated and 80% be handled by a human. But if you think about an agentic system, 80% are being handled by the AI agent and 20% human or maybe more. Right. So the overall cost of the system might be going down if.

[Dr. Darren]

Yeah, yeah.

[Sid Brott]

You're spending less on human labor. Right. So the agentic system could.

[Dr. Darren]

Yeah. This sounds like a true. People are starting to use the word hybrid workforce. I think heterogeneous workforce is a better word because I've got humans, I've got agentic flows, and I've got deterministic workflows, things that happen and they're always going to happen the same way, no matter what. You know, once it's triggered, it does it. So maybe this is a heterogeneous workflow because I'm using more than two a lot of times. I think we think the hybrid is like two different types of things. I'm using everything to get this done. This is actually pretty cool.

[Sid Brott]

Yeah, exactly. And it's like they work. They all work together in the same system. So it's about designing that system in terms of, like, you know, what is. What are the inputs and what outputs do you want? And then how do the. Which parts can you fully automate? Just like with a deterministic thing, which ones require an agentic thing and which ones require a human. Now we are moving to workflows where the agentic is like, it's almost 100% agentic and the human is there just to review. So I'll give you another example. Same case of like Nike shoes. Let's say a customer writes in to say, I want a refund on the shoe right now. In a deterministic workflow, maybe you would have just said, for any questions related to refunds, send it to Hume, because you can't determine what the threshold is and all of that kind of stuff.

[Dr. Darren]

Right.

[Sid Brott]

The agentic workflow, the agent can then determine, okay, what is the price of the shoe? How much is the refund? What is the value of this customer? Have they spent tens of thousands on us? Are they an influencer or are they a first timer? Are they a repeat refunder? Right. The agent can go check all of those different things and, and make a choice based on that and say, okay, I think this person merits a refund or maybe merits 70% refund or whatever, or a coupon code for the next purchase and then passes it to a human just for a quick judgment call. Hey, does this seem right to you? And then the human says yes or no? And then it just goes and does the thing.

[Dr. Darren]

That's very cool because the agent is doing all of the research for the human to make the decision. Yes.

[Sid Brott]

Yeah, it does. It can even come up with a suggested decision, like, hey, doing all the research. Here's what I found. Here's a suggested decision. You know, it's okay to give them a 50% discount or refund. This falls within the boundaries that we've set as a company for refunds or whatever. What do you think? Yes or no?

[Emergent Narrator]

Quick break and a simple idea. Most organizations don't fail because of strategy. They fail because execution breaks down. That's where emerging comes in. They help organizations build what they call a work operating system. A way to align people, process, and technology so ideas actually turn into outcomes. Not theory, not transformation theater. Real systems that drive real results. If you want your organization to execute better, not just plan better, check out immersion.com.

[Dr. Darren]

This is. This is really, really cool. So my next question. Because my question goes directly to. All right, so there's got to be something orchestrating between these three different types of resources, the human, the deterministic, and the agentic flow, there's something orchestrating this. Is that an AI orchestrator, or what you call that thing that takes the initial, hey, I got the initial request or whatever it is, right?

[Sid Brott]

Yeah, yeah. It could just be like an AI orchestrator. It could be like a very basic AI classification that just looks at what's coming in and classifies whether, hey, this can just go through a determinist workflow. This needs to go to humans, very complex. Or this can go through the agentic workflow and get processed that way. Right?

[Dr. Darren]

That's very cool. All right, now let's talk about the LLMs on the back end on this. Do I have to use a public generative AI to do this work? Or are there open models that I can use? Because I might have some information that I really can't let go out to a public Gen AI Because I'm thinking like patient, we could use a system like this to manage patients at a hospital. I, I can't send out, you know, personal patient information. I can't send that out to a public Gen AI. That's kind of breaking hipaa, right?

[Sid Brott]

So yes and no. I mean like with, with your companies like OpenAI, Anthropic, Gemini, Google Gemini, they do have those HIPAA certs and so on and they do like if you get the enterprise version then they do say that your data is secure, it won't be trained on, it's going to be on servers located in the U.S. whatever. Right now it's the same thing as saying using Google as your email provider, if you're a healthcare provider. Right. So your data is still living on Google servers. It's all your email and patient data that's going through emails. But Google is not like selling that data or training on the data or anything. So they're, that's kind of like your data is still safe in that way. And so if you're sending it to a language model for a response and a language model sits on Google servers in the US or OpenAI servers and OpenAI said you're part of our enterprise account deal, which means it's not going to ever be used in training or whatever, then it's sort of like you're still compliant. That being said, there are open source models that are as good as, almost as good as current gen, state of the art paid models which you can host on your own server. So if you do have your own servers internally, which a lot of enterprise companies do, you can run a, an open source model that is made by Meta or some of the other open source labs that have created. Even Google has their own open source models. There's one called Gemma which is like Gemini, but Gemma is a good model.

[Dr. Darren]

Yeah, I've used that one.

[Sid Brott]

Great model. You of course have all of those open source Chinese models that are actually.

[Dr. Darren]

Oh yeah, yeah. I trust those 100% by the way. Not really.

[Sid Brott]

All that is to say the open

[Dr. Darren]

source connect from the Internet when you use those and see what happens.

[Sid Brott]

Yeah. All of that is to say the open source technology is nearly as good as the paid.

[Dr. Darren]

Okay, so that is an option if I'm worried about my data loss and let's say I'm in a stealth startup, right? Yeah, I don't, you know, I'm very cautious about my data privacy. There are options for me to do these same kinds of agentic flows with private gen AI, Absolutely, yeah. Oh, cool. Very, very, very cool. All right, so, yeah, I've got agentic. I think I understand agentic now. Right, because it's. It's doing stuff. It can actually do things for me, too. That's the big difference between it just sitting there, just giving me a response. It's querying other information. It could be reserving inventory for that new sneaker that it's planning on, on sending out to the person. It could be sending shipping labels out. It could do all this stuff for me.

[Sid Brott]

Right, That's a very good point. So, like, that's the difference between your chatbot way of working, which was maybe last year. You know, AI moves so fast. I didn't think that happened last year.

[Dr. Darren]

Last year, Sid. That's like forever ago, last year. Right.

[Sid Brott]

But if you think about, like, how you use. If you're using ChatGPT, if anyone listening to the show right now is you're using ChatGPT as your regular thing, and you find yourself going and looking for data and uploading it into ChatGPT and asking ChatGPT for an answer. And ChatGPT tells you what to do, then you go and do the things. Think about what that means. It means that you're doing all the work to collect the context, and the AI is telling you what to do. So the AI is telling you what to do, and you, the human, are doing the work. But if you think about the agentic way of working, which is what we have just been talking about, what's happening is the AI is doing the work. The AI is collecting the data, and then asking you, hey, is this right or wrong? And you are telling it, yes, go do the thing. And then it goes and does the thing. So now we have flipped the model. Right?

[Dr. Darren]

Okay. I love that, Sid, because you're putting the. The AI overlords at bay and you're switching the tables on them. Right? We're in charge with the Gentex flows. We're in charge, and we need to make sure that we stay in charge. I love that.

[Sid Brott]

Yes. Yeah, exactly.

[Dr. Darren]

Yeah, that's awesome.

[Sid Brott]

I think, like, when I. When I talk to people at companies as well, and we go in and we do these AI audits to really understand how are they working, where's all the manual work going, and where can we build AI agents for them? And our goal is to move them into this agentic workflow because a lot of them are still stuck in that ChatGPT kind of workflow where they are sort of like trying to build the Perfect prompt and try to get all the data and like ask ChatGPT and have and do what ChatGPT says. And so it's very eye opening when I frame it this way as instead of the AI telling you what to do and you doing all the work, you tell the AI what to do. The AI does the work. That's the easiest way of working.

[Emergent Narrator]

One more quick note. It's easy to talk about transformation, it's harder to prove it. Emergent has been doing this for over 17 years, working with organizations like GSK, SAP, Lloyds Banking Group and JB Hunt. But what matters isn't the logos, it's the outcome, better ways of working, stronger teams, technology that actually serves the business because they're not focused on short term wins, they're focused on building something that lasts. Get ahead, stay ahead. Visit emergen.com Sid what's the hardest part

[Dr. Darren]

of going into an organization and dropping an agentic flow in? Can I just drop it in and your AI goes crazy and starts finding things or you know what, walk me through that process. What's the hardest part?

[Sid Brott]

Yeah, I mean I wish we had that, that could go ahead and do that and you know, like that would probably be one of our goals to build eventually. But I think right now the biggest, the biggest problem is the fact that especially at larger companies where there's so many different teams, so many different data silos, you know, like the tools as well, so many different tools that don't talk to each other, being able to first get all of that data into a common data context layer with which we can build agents on top. That's the biggest problem and that is solved right now by us just going into the company and really deeply understanding the way that they work and how they work, what tools they use and so on so that we can get a full picture of that workflow map and then build that for them.

[Dr. Darren]

Do you ever run into data access problems with, you know, data security and things like that because you are operating on behalf of somebody else, these agentic flows. So do you ever run into kind of that data permission or execution type of permission issues or is that pretty straightforward to do nowadays?

[Sid Brott]

That's straightforward to do especially when we work with the client to, we'll get access to the assistance. So we do have to go through their sort of data security protocols and data access protocols. But we'll, we build the tools and we build the agents into their existing servers and so on. Right. So it sits within their, with their data, within their servers and their Architecture, and, you know, that's kind of like the. Also the easiest way to do it.

[Dr. Darren]

Okay, so these. These agents are actually running on their infrastructure, whatever it is, whether it's in the cloud or on prem or on a laptop even, right? You can run these agents on laptops?

[Sid Brott]

Yeah, absolutely.

[Dr. Darren]

So that's really slick. So you're not like a full SaaS solution where, oh, I'm now relegating all my data to another provider that's talking to another SaaS provider that's. And who knows where my data is. Now you guys go to where the data is and where the work is happening. I really like that.

[Sid Brott]

I mean, some of these companies, they will have, like, these sort of data lakes, right? And they have their own, like, databases and such. And then we could just work off of that on their architecture and their servers, and that's much easier

[Dr. Darren]

now. That's very cool. All right, so the next big question I have for you, and now's the time you get to pitch your company. If I. If I want to get involved, if I want to do this, what are the steps I need to take? You know, let's say I want. Hey, I want Sid, I want your company to come in and help me automate my. My processes for producing my podcast and advertise and all this stuff around it. Because there's a lot of steps. People don't know that about podcasts. All right, what. What do I do? I just call you up and say, get over here, Sid. Come to California, where it's nice and warm.

[Sid Brott]

And I mean, you know, I was just there. I was in la and I came back yesterday to Vancouver. It was very nice and warm. But, yeah, I mean, that's basically it. You know, usually, like, someone in the company will reach out to us and say, hey, look, you know, we're scaling, we're growing, but XYZ process takes two weeks. Or this thing is a bottleneck, or, you know, we're being held back by our processes and tools. Can you come and figure out what's going on? Oftentimes, these companies don't really know what the problem is, but they know there is a problem. They know that for them to scale, they have to maybe double their workforce, which is, you know, not what they want to do. So they're like, is there a way for us to scale without doubling the workforce? Or that a critical part piece of their business flow is being bottlenecked by something? And so they're like, okay, there's some bottleneck here. Can you help us Figure out what that is and see if we can solve it. And so like, part of what we do is also just uncover what the problems are and put a number to it and then sort of just show you. Okay, here's.

[Dr. Darren]

You know, it sounds like you're a good old fashioned process consultant. You know, that's what you are.

[Sid Brott]

That's basically it. I mean, I started my career in, at Deloitte, at consulting. And this is what we used to do. And I think back and I'm like, this is basically the same thing, except obviously I've got this massive AI expertise and so I'm able to look.

[Dr. Darren]

Yeah, now you don't, now you don't leave behind a set of people sitting at desks doing this. You leave behind some agentic flows for them, which is basically.

[Sid Brott]

Yeah, yeah. Because like, if you think about what, what is what Deloitte does and so on, they'll come in and they'll build custom software and then they'll bill you, you know, $100,000 a month to maintain their software and to keep making new changes and such. Right. And to train your team on it. But now if you, if you could just build AI agents for a fraction of the cost that just like autonomously keeps working, then, you know, you don't really need all of that stuff.

[Dr. Darren]

Well, you brought up the T word, the training, the training word. How much training do you have to do? Because I'm assuming you guys deploy these custom AI orchestrators and workload and work agentic flows with them. Is there training that you do with them so that they can now maintain or add new agents into the flows and things like that? Or is, is, is it, hey, don't touch our stuff. We'll come and create new agents for you. How, how do you guys work that issue?

[Sid Brott]

Yeah, it's more like we'll train them on how to use it and how to work with the agents. You know, kind of like what it can do, what it can't do, or like what it's not meant to do. And so how do you, how do you interact? And we can, we will, if they want to know how to like extend the agent or add or customize it, then sure. But oftentimes it still requires some technical ability and technical knowledge to come do that. So, you know, they'll, they'll come back to us if they really need like major changes. But yeah, like if, if the basic training of just like, how do you work with it, that kind of like, if we've, if we have understood the process well enough then, and we've solved it with the agent, then there's really not much else to do. Right.

[Dr. Darren]

Oh, that. That's super cool. You know, this could be some really interesting new business models, because if you think what you're giving them is resources, you're giving them resources to help them do their jobs more effectively. And in the old model, right, with Deloitte or Kelly Services, you're actually renting butts and seats, basically. Right. You're renting resources. And if you want to actually not pay monthly for that agent, you can buy that resource by hiring them, but you have to pay a fee. This could be a very interesting type of. Type of business model that no one's ever really thought of. Could you imagine your. Your. Your business having a. A stable of AI agents that you rent out to people? That would be kind of cool.

[Sid Brott]

Yeah. I mean, that's. That's one. That's kind of like One Direction. We were also thinking of, you know, because. Yes. Yeah. Like, if we can. The thing with these AI agents is, like, they do need to be customized to your business and your workflow, and every business is your.

[Dr. Darren]

Yeah, that's true. That is true. So they're not very repeated.

[Sid Brott]

Right. But. But there are some core components that we're trying to build that. That we can repeat. And so it becomes like a productized service where we say, okay, like, we have this sort of agent that we use for building decks, right, like pitch decks or whatever. And so for any. Any company or any team that has to create a lot of pitch decks, like a sales team or an investment team, you know, we could still use the same core tech, the agentic workflow, but we could just come in and come into your organization and customize it based on your tools and your business.

[Dr. Darren]

This is super cool. This is super cool, Sid. All right, so if people want to reach out to you, where do they go?

[Sid Brott]

Just go to refoundai.com r e f O-U-N-A I.com and you'll find all the information you need there.

[Dr. Darren]

Oh, this. This is awesome. Sid. Thanks for coming on the show. This has been wonderful. I learned some really. Some really cool things. I'm going to start applying myself. My head was, like, going, how am I going to automate my podcast? How am I going to. So I will always be behind the mic. Yeah, let's just put it that way.

[Sid Brott]

I'm not going to do a deep fake. I do think that I'll give you a starting exercise if you Want to.

[Dr. Darren]

Okay, give me a starting exercise.

[Sid Brott]

Like you're trying to automate your whole, I mean, parts of your podcast workflow. So what you should be doing is I'm sure you have other people on your team and your various tools map out the whole process from start to finish, from how do you find a guest speaker all the way to how does that podcast get published online and promoted. Right. Every step of the way, map it out. See right now who's in charge of what, what tools are along the way and what's taking the most time. Right now. There are some things that an AI agent shouldn't be doing, like being the

[Dr. Darren]

host of the podcast that can't be

[Sid Brott]

me, the human domain. So you have to mark those off as like, okay, this must stay human. This must stay human. But everything else that could become AI agent, right? If it is like doing research on the guest, you don't really need to go into Google and LinkedIn and do it. Your AI agent can go and pull up a full profile if it is publishing something to social media. Once you have the final recording again, an agent can take that, pick that up, turn that into social media content in your voice and tone, and schedule it for you. Right. And so you can then decide after you map out the full workflow, what should be human, what can go into AI. And you know, especially looking at the parts that take the longest or most repetitive, those are the ones that are the lowest timing fruit to automate.

[Dr. Darren]

All right, I'm going to do that, Sid, and I re. I will report back.

[Sid Brott]

Yeah, you send me an email with like the whole thing and let me know what he's and I can use my feedback.

[Emergent Narrator]

Yeah.

[Dr. Darren]

Awesome. Hey, thanks again, Sid, for coming on the show.

[Sid Brott]

Awesome. Thank you so much, dad. And it was a pleasure.

[Dr. Darren]

Thanks for listening to Embracing Digital Transformation. If you enjoyed today's conversation, give us five stars on your favorite podcasting app or on YouTube. It really helps others discover the show.

[Emergent Narrator]

If you want to go deeper, join

[Dr. Darren]

our exclusive [e-mail] embracingdigital where we share bonus content and you can always connect with other change makers like yourself. You can always find more [e-mail] until next time, keep embracing the digital transformation.